[SOLVED] fsck pass

Hi, this may be an odd question but I've only just noticed this, as I'm re-installing to my laptop after a long period of disuse during the final term of uni. In the setup guide I've read:
<pass>
used by the fsck(8) program to determine the order in which filesystem checks are done at boot time. The root filesystem should be specified with a <pass> of 1, and other filesystems should have a <pass> of 2 or 0. Filesystems within a drive will be checked sequentially, but filesystems on different drives will be checked at the same time to utilize parallelism available in the hardware. If the sixth field is not present or zero, a value of zero is returned and fsck will assume that the filesystem does not need to be checked.
In my fstab, all the partitions have been given a <pass> of 1. Do I need to change this or worry about it. If my assumptions on it are true then it simply mean every partition is checked with fsck.
Thanks,
Ben9250.
Last edited by Ben9250 (2011-06-17 02:33:32)

I recommend that you change it. I doubt your PC will explode if you don't, but there's probably a good reason why only the root filesystem is supposed to have a priority of 1. Partitions will still get checked if they have a pass priority of 2.

Similar Messages

[SOLVED]fsck fails on bootup due to other hdd name

Hi,
is set up arch today on an initially 2 hdd sys ...
after it was set up and the files i wanted to keep were moved to the hdd which would remain, i disconnected one hdd ..
(i had to change grub/menu.lst sdb1 ---> sda1 due to this disconnect)
unfortunatly during bootup fsck still thinks it must check sdb1 which will fail because its not there
How do i solve this,
thanks,
Last edited by birdspider (2008-01-14 18:35:56)

ok sry, I had just to remount the hdd RW and rename those issues in fstab

How to solve parameter passing problem?

I have a report that can be place on any Portal page and it requires several parameters.
The report is based on a query:
select bldgname
from building
where (bldcntry = :cntry or :cntry is null) and
(bldstate = :state or :state is null) and
(bldcity = :city or :city is null)
The LOVs for the bind variables need to be linked, i.e. State LOV is function of Cntry LOV and city LOV is function of both cntry LOV and state LOV.
I can't use the Customization screen because it doesn't allow linked LOVs.
I can't use a master detail form because I would have to create a table consisting of the 3 bind variables, M/D forms don't work with views. Since this is just one of 60 reports I have to do, it would mean adding 60 of these pseudo tables.
I was thinking of placing a form, which can have linked LOVs, on the same page as the report, and then calling the page and passing the values of the LOV to three page parameters. Unfortunately, I need to dynamically determine the url of the page the report and form are on and p_page_url doesn't seem to be accessible from a form, although it is from a report, I have a TAR on this issue.
So I wonder if anyone has an idea how to solve this problem?
Larry

Hi,
Try calling the report from the success procedure of the form
Here is an example which calls the report from the success procedure of the form.
declare
flightno number;
ticketno varchar2(30);
blk varchar2(10) := 'DEFAULT';
l_url varchar2(2000);
begin
flightno := p_session.get_value_as_varchar2(
p_block_name => blk,
p_attribute_name => 'A_FLIGHT_NO');
ticketno := p_session.get_value_as_varchar2(
p_block_name => blk,
p_attribute_name => 'A_TICKET_NO');
l_url := 'http://domain/pls/reportname.show?p_arg_names=flightno&p_arg_values='||flightno||'&p_arg_names=ticketno&p_arg_values='||ticketno;
call(l_url);
end;
Thanks,
Sharmila

[Solved] Fsck Maximum mount count set to -1

I just noticed that fsck has not been running on my system and found the following:
dumpe2fs -h /dev/sda2 | grep -i 'mount count'
dumpe2fs 1.42.12 (29-Aug-2014)
Mount count: 110
Maximum mount count: -1
and
dumpe2fs -h /dev/sda3 | grep -i 'mount count'
dumpe2fs 1.42.12 (29-Aug-2014)
Mount count: 110
Maximum mount count: -1
My fstab is as follows:
/ ext4 rw,noatime,data=ordered,discard 0 1
/home ext4 rw,noatime,data=ordered,discard 0 2
My understanding is that the "-1" for the maximum mount count means the file system doesn't get checked at all.
What would cause the maximum mount count to be set to -1?
One thing that I did do in the past was to remove the fsck hook from mkinitcpio and copy the appropriate systemd files per the wiki on quiet boot.
Last edited by Gumper (2014-11-23 15:01:28)

I marked this post as solved, but I'm seeing this when I check my boot partition.
dumpe2fs -h /dev/sda1 | grep -i 'mount count'
dumpe2fs 1.42.12 (29-Aug-2014)
dumpe2fs: Bad magic number in super-block while trying to open /dev/sda1
NAME FSTYPE SIZE LABEL
sda 477G
├─sda1 vfat 512M
├─sda2 ext4 25G
└─sda3 ext4 451.4G
Do I need to be concerned with this, or is this normal for an EFI system boot partition?

[Unknowingly Solved] Bash, passing password to ssh-add

Hey guys,
I'm working on a script to add my ssh key to my ssh-agent using pam_exec.so. pam_exec passes the password to my script over STDIN, which is fine. But I can't for the life of me figure out how to pass the password to ssh-add.
I've already seen this question, but would really perfer not to have to install extra/expects just for this on script.
Any ideas?
Last edited by EvanPurkhiser (2013-08-14 05:40:34)

Wow.. Ok. So I feel dumb. I was way over thinking this one.
Since pam_exec passes the password to the script over STDIN (WITH a trailing null character) I actually just have to add the ssh-add command somewhere in the script and it will read from STDIN.
I was actually stuck on this for awhile and had already tried
echo "myPassword" | add-pass
which obviously didn't work. And even now, trying something like this
printf "MyPassword\x00" | ssh-add
still doesn't work.
I'm actually a little perplexed as to WHY this worked.
Here's the full script
#!/bin/sh
# Takes a password from STDIN, starts the ssh-agent as a systemd user service,
# and decrypts the ssh key using the provided password, adding it to the agent.
# Handle inital checks as root
if [ $(id -u) = 0 ]
then
# Don't execute if the user-session isn't running
systemctl -q is-active user-session@${PAM_USER} || exit 0
# Re-execute this script as the user to add their key (while piping STDIN)
cat | exec su ${PAM_USER} -c $(realpath ${BASH_SOURCE[0]})
# Handle adding the key as the user
else
# We need to specify the XDG_RUNTIME_DIR because pam_systemd won't have run
export XDG_RUNTIME_DIR=/run/user/$(id -u)
# Get the SSH_AUTH_SOCK variable from the user session
export $(systemctl --user show-environment | grep SSH_AUTH_SOCK)
# Ensure the ssh-agent service is started
systemctl --user start ssh-agent
ssh-add
fi
... Marking as unknowingly solved
Last edited by EvanPurkhiser (2013-08-14 05:40:58)

[SOLVED]fsck hook with new mkinitcpio.conf

Last night when I upgraded mkinitcpio it provided a new mkinitcpio.conf file as a pacnew. When comparing this new file to my previous config file the only difference I found was that the new one included the fsck hook. From what I've read it says that this hook is strongly recommended but only gives a reason for it if you have a separate /usr partition, which I do not. I have separate /boot, /swap, /, and /home partitions. Should I include this hook? Does it hurt? Doesn't matter?
Last edited by Lexworth (2012-03-15 14:52:47)

karol wrote:
You should have checked the wiki https://wiki.archlinux.org/index.php/Mkinitcpio#HOOKS too:
Adds the fsck binary and filesystem specific helpers. If added after the autodetect hook, only the helper specific to your root filesystem will be added. Usage of this hook is strongly recommended, and it is required with a separate /usr partition.
I did check the wiki. I mentioned what I had gathered from my reading in my original post, perhaps you should have read it. I then went on to explain that this didn't give much indication for including the fsck hook without a separate /usr partition. So I came here (Newbie Corner) to ask if someone could help me with my understanding, which Gcool was so kind enough to do.

[Solved] Assembly - Passing Integer Command Line Parameters

This is as good a place to ask as any I suppose, so here goes.
I'm really only interested in knowing if it is at all possible to push actual values, in this case integers, on the stack from the command Line, i.e. passing arguments and if so how, because as far as I can gather, it is simply impossible to pass anything other than strings.
Yes I am a noob at assembly, if anyone should doubt that.
Best regards.
I thought the best thing to do is simply to explain what I'm actually doing.
In short, I need to derive the Lightness/Luminance level of a pixel. This of course can be done a number of way. Initially I experimented with image magic, but when I couldn't figure it out, I thought: "Hey, why not learn something along the way?"
So here I an, trying to do some assembly under linux.
For those interested the process is actually rather sinbple:
1. get and convert the rgb input as needed.
2. Determine the highest and lowest channel (progress so far):
mov ax, $1;
mov cx, $2;
mov dx, $3;
cmp ax, cx;
jng else;
mov ax, ax;
mox bx, cx;
jmp next;
else:
mov ax, cx;
mov bx, ax;
next:
cmp dx, ax;
jng else2;
mov ax, dx;
jmp end;
else2:
cmp bx, dx;
jnl end;
mov bx, dx
end:
3. add the two values and divide by two. possible implement a way to check if the addition is even or not, so to enable proper rounding.
4. output the result in one form or another.
Last edited by zacariaz (2012-09-03 20:30:14)

Lux Perpetua wrote:
zacariaz wrote:
Well, at the same time it makes sense and it don't.
I makes sense to have a "common language", so to speak, so that various applications can be piped together.
At the same time however, it's a waste to have the majority of the code dedicated to making conversions. First one for the input and then one for the output.
But that life I guess.
Thanks for now.
When you run a program on a Linux system, it ultimately all boils down to the following function:
int execve(const char *path, char *const argv[], char *const envp[]);
Without going into full detail: the first argument is the path to the program being executed. The second argument is literally the "argv" argument list the program sees (which is a null-pointer-terminated array of pointers to null-character-terminated strings). The third argument is the environment. Relevant to you is the second argument. It doesn't matter if it's an assembly program, a C program, or a shell script; the "command-line" arguments to the program, i.e., the argv array, are pointers to strings.
Lest you have the brilliant idea to take an integer apart into bytes and stick them in a string as one of the argument as some others have suggested, you should consider the following: what if one of the bytes of the integer is 0? That technically ends the "string," and everything following that byte is discarded by execve. Adding 1 to all the bytes wouldn't work, since 0xff would wrap around to 0 again. To make it work, you'd need to encode the integer in some other complicated way to avoid null characters and then decode it inside your program. In other words, you might as well just pass an ordinary string and parse it with atoi or strtol.
You are of course right, and thanks for the very nice explanation, however it's still a pain in the but.
I like working with number, not with letters, but enough of this.
Thanks to all and to all a good night, day or whatever you please.
Best regards

[solved] fsck and encrypted /home

Hello,
After testing arch for some weeks, I made the "final" switch this week-end: I moved my LUKS encrypted /home from ubuntu tu arch. During this process I had a problem with fsck on my encrypted /home. I put 0 in /etc/stab to disable fsck:
/dev/mapper/home /home ext3 defaults 0 0
So my question is: is it possible to activate fsck on my /home at boot time ? Or should I let this as it is ?
Thanks,
Last edited by boulde (2009-10-06 07:53:06)

I don't remember exactly but there was fsck in the error message during the boot, it was not possible to boot. But I have just tested again with fsck and it works well !
So this error was maybe not due to fsck (I had various issues with crypttab options, qwerty/azerty passphrase, etc.) ... but I remember that before deactivating fsck the system was not booting, after it was.
Anyway, now it works :-)

[Solved] fsck.vfat isnt installed

I was going to run an fsck on my thumbdrive but for some reason fsck.vfat isnt installed, yet all the others are (ext2-4, jfs, xfs, etc...) and I have no idea on where to get it from. Can someone tell me?
Last edited by brando56894 (2009-06-22 20:50:09)

Install dosfstools.
Last edited by slyson (2009-06-22 18:56:50)

How to pass jv for ED missing in excise invoice

Hi Experts
Could you pls tell me how to pass ED( Excise Duty ) entry through JV ( J1ih).
Case: When the outgoing Excise Invoice was created one of the excise Duty was missing.
Now i want to pass the ED against that invoice through JV .
N pls also suggest how to assure that the jv passed correctly against the Previous invoice.

NOW SOLVED BY PASSING jv THROUGH j1IH.

Possible to automount AEBS HD

My configuration includes an iMac running Mac OS 10.7.5, an Apple Extreme Base Station
with a partitioned USB hard drive connected to it, and various other machines wirelessy connected. I am
trying to configure the hard drive so that its volumes are accessible with different permissions
depending on the user. I want this configuration to survive a restart and wake scenario.
It appears to me that the only configuration workable with Lion and networked drives is for open
(Read Write access) access for all users that mount the drive (or any of its volumes). This involves
simply accessing the server via command-K and supplying the password.
A permanent configuration via some automount solution has been raised in a number of queries
and I have tried all the solutions to no avail. An example of a more comprehensive thread is:-
https://discussions.apple.com/thread/3221944?start=0&tstart=0
I have tried using the fstab and automount solutions (using a mount map file) but they both result in
root mounted volumes. It seems some regression was introduced in Lion that has broken the auto-mounting of
networked drives (note, mounting the drives using mount_afp works fine).
I wonder is anyone aware of a solution and more importantly, for Mountain Lion users, does this problem still exist ?
If I upgrade to ML will this allow an auto-mount configuration?
Notes
====
For anyone with the time to try on ML, the fstab mods are of the form (man fstab):-
<Id of Vol. to mount> <Mount Point> <fstype> <options> <dump param> <fsck pass no.>
192.168.0.3:/Iomega-100GB /Volumes/myIomega-100GB url rw,auto,automounted,url=="afp://username:[email protected]/Iomega-100GB/" 0 0
Where username is login name and password is the disk password set via AirPort Utility.
Mods to fstab can be tested via sudo automount -vc (avoids reboot etc.). Can then delete/reset fstab and re-run this command to
get system back to previous state.
Console application will show some errors and info.
From man documentation and console output, the above fstab mod leads to the invocation of mount_afp.
I can see that there are potential issues with this approach. Such as, how does the username/pw apply to other users?
Given this mount will take place during a boot then volumes could at most be potentially available to users
with the final stage of credential checking only taking place at the point of use of volumes. This would explain
why mount_afp works (there are no ambiguities of who the user is).
From the automount documentation I have noticed that there is an 'executable map' option that allows conditional
mounting of volumes. I have not explored this as I don't understand the root mount behaviour (where this executable
option is likely to lead!).

Thanks VikingOSX. The Apple info is basic and does not cater for giving each user different permissions to the volumes (as you would normally expect from a standard ownership model .. in fact, the behaviour you get when you plug the drive directly into the machine). I want to be able to mount a drive as a network drive such that its ownership is respected. I can see that this would entail some form of matching users across a network in order to establish the 'owner'.
I have tried dragging drive icon into the account login items however, this does not address file ownership. It just provides open access to that user. I have read in some threads that MacOS X works on the basis of a floating ownership whereby whoever mounts the drive gets R/W access to all files. This looks like what I am seeing.
I am beginning to think the type of functionality I am looking for is more in the realms of OSX Server ?

Human tasks, complex types and variable assignment.

Hi folks,
I encountered a problem while working with 10.1.3.1 which I haven't managed to solve yet and would be grateful for assistance. I have defined a complex data type (just a sequence of strings) and want to fill that data type by going through a flow of screens, e.g. through several human tasks. I have defined that type as a variable to the BPEL process and have managed to assign simple expressions to the variable and pass that into a Human Task. In the task I see now a simple input form for all attributes of the type.
Now the problem. After I fill out the form and complete the human task, I have
The global variable:
<outputVariable>
     <part xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" name="payload">
          <MyProcessResponse xmlns="http://xmlns.oracle.com/VacationRequest">
               <vorname/>
               <nachname/>
               <strasse/>
               <nummer/>
               <postleitzahl/>
               <stadt/>
               <maximaleDauer/>
               <minimaleDauer/>
          </MyProcessResponse>
     </part>
</outputVariable>
and the return value from the human task:
<task>
     <title>CaptureData</title>-
          <payload>
               <MyProcessResponse>
                    <vorname>David</vorname>
                    <nachname>Beckham</nachname>
                    <strasse>HighStreet</strasse>
                    <nummer/>
                    <postleitzahl/>
                    <stadt/>
                    <maximaleDauer/>
                    <minimaleDauer/>
               </MyProcessResponse>
          </payload>
</task>
Now the only thing I would like to do is to replace the original values in the variable with the new values that have been returned from the task, this should be performed in an extra assign step. Sounds simple. JDeveloper wouldn't even let me klick through the return value from the HumanTask, so it seems as if I have to do things manually.
These are my tries:
Command 1:
<copy>
     <from variable="CaptureData_1_globalVariable"
part="payload" query="/task:task/task:payload" />
<to variable="outputVariable" part="payload"
query="/ns1:MyProcessResponse"/>
</copy>
Result 1:
<outputVariable>
     <part xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" name="payload">
          <ns0:MyProcessResponse xmlns="http://xmlns.oracle.com/bpel/workflow/task"     xmlns:ns0="http://xmlns.oracle.com/VacationRequest">
               <MyProcessResponse xmlns="http://xmlns.oracle.com/VacationRequest">
                    <vorname>David</vorname>
                    <nachname>Beckham</nachname>
                    <strasse>HighStreet</strasse>
                    <nummer/>
                    <postleitzahl/>
                    <stadt/>
                    <maximaleDauer/>
                    <minimaleDauer/>
               </MyProcessResponse>
          </ns0:MyProcessResponse>
     </part>
</outputVariable>
--> There is one MyProcessResponse element too many
Command 2:
<copy>
     <from variable="CaptureData_1_globalVariable"
part="payload" query="/task:task/task:payload"/>
<to variable="outputVariable" part="payload"
query="/ns1:MyProcessResponse/ns1:vorname"/>
</copy>
Result 2:
<outputVariable>
     <part xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" name="payload">
          <MyProcessResponse xmlns="http://xmlns.oracle.com/VacationRequest">
               <ns0:vorname xmlns="http://xmlns.oracle.com/bpel/workflow/task"                     xmlns:ns0="http://xmlns.oracle.com/VacationRequest">
                    <MyProcessResponse xmlns="http://xmlns.oracle.com/VacationRequest">
                         <vorname>David</vorname>
                         <nachname>Beckham</nachname>
                         <strasse>HighStreet</strasse>
                         <nummer/>
                         <postleitzahl/>
                         <stadt/>
                         <maximaleDauer/>
                         <minimaleDauer/>
                    </MyProcessResponse>
               </ns0:vorname>
               <nachname/>
               <strasse/>
               <nummer/>
               <postleitzahl/>
               <stadt/>
               <maximaleDauer/>
               <minimaleDauer/>
          </MyProcessResponse>
     </part>
</outputVariable>
--> This was moreless expected, since I tried to copy the stuff to some wrong place.
Command 3:
<copy>
     <from variable="CaptureData_1_globalVariable"
part="payload" query="/task:task/task:payload/ns0:MyProcessResponse"/>
<to variable="outputVariable" part="payload"
query="/ns1:MyProcessResponse"/>
</copy>
Result 3:
--> Task does't return (!!!). In the BPEL Worklist, the task is completed, however in the BPEL console it still apears as "waiting to return from Human task".
Does anyone have a hint on how to copy the values so that I get exactly the structure that I printed at the very top?
Thanks for help. Rock On !

Hello everyone,
It seems that we are talking about different things, I am afraid. Let me describe the scenario once again:
- I want to have a BPEL process with more than one role involved
- This process should be used to gather some complex data.
- All data gathering should be done through HumanTasks.
So I defined an xsd complex type, that contains a sequence of fields. This complex type should be part of the process payload.
Now before HumanTask 1 is called, I copy the contents of this variable into the input for tasks1. This works fine and I see the appropriate values, when opening the Human Task in the BPEL worklist application.
Now after HumanTask1 is completed, the values I entered inside that tasks are returned as a variable from the HumanTask. My problem is, that I cannot copy the contents of the task back into the main process. With simpleTypes this is relatively ok, since Jdev supports the assembling of the copy operation, however for complex tasks it fails.
A typical scenario for this would be where e.g. a customer fills a shipping address and afterwards a store staff fills product details in the same order. In Java this could be solved via "pass by reference". Who can provide an example?

CS193p 2010-11 Fall, Assignment 3

TL;DR: How to let a class method of a view communicate with its delegate, owner or so?
The assignment is to iterate over pixels to draw points over X and Y axes. The X axis is the X, the Y axis is the evaluation of the expression of X, sent by a view controller called CalulatorViewController.
Here's more explanation of what I did. CalculatorViewController has a model called CalculatorBrain and a .xib view. It's a normal calculator with a button called "solve" which passes the expression to the brain, with a dictionary of the values of the variables. 2 for X for example.
Therefore the brain will evaluate X + 2 to be 4 and return the result, which is 4.
Another View Controller is the graph one, which is called AxesDrawerViewController. It has a model which is called AxesDrawerBrain and a view which is called AxesDrawerView, which has another class called AxesDrawer which works as a core graphics class.
The model is reponsible for returning the appropriate scale when Zoom In or Zoom Out buttons are pressed. The calculator simply tells the model that it changed by X rate, then asks it for the scale, then passes it to AxesDrawerView which passes it to AxesDrawer, it draws, and boom, it's there.
There's only 1 problem which is actual drawing of the graph of points over pixels. Therefore I allocated AxesDrawerViewController at the initialization of CalculatorViewController, set its expression, and set its delegate.
The expression is what is recorded with each button pressed. For example I pressed 5 then + then X, so the expression will be 5 + X.
At the initialization of AxesDrawerViewController, I set the AxesDrawerView's delegate to be the AxesDrawerViewController. Therefore I can simply get the delegate through AxesDrawerView from the delegate by adding a method to AxesDrawerViewController which asks its delegate, which is CalculatorViewController to return the expression. Then I make an instance of CalculatorBrain, send it the expression, and the dictionary with the value of X I'm iterating over, so I can return it to AxesDrawer whenever it asks me to using a delegate.
The problem is that AxesDrawer's methods are class methods, so it can't see delegates. I don't know how to bypass this problem.
Any help is appreciated. Thanks.

TL;DR: How to let a class method of a view communicate with its delegate, owner or so?
The assignment is to iterate over pixels to draw points over X and Y axes. The X axis is the X, the Y axis is the evaluation of the expression of X, sent by a view controller called CalulatorViewController.
Here's more explanation of what I did. CalculatorViewController has a model called CalculatorBrain and a .xib view. It's a normal calculator with a button called "solve" which passes the expression to the brain, with a dictionary of the values of the variables. 2 for X for example.
Therefore the brain will evaluate X + 2 to be 4 and return the result, which is 4.
Another View Controller is the graph one, which is called AxesDrawerViewController. It has a model which is called AxesDrawerBrain and a view which is called AxesDrawerView, which has another class called AxesDrawer which works as a core graphics class.
The model is reponsible for returning the appropriate scale when Zoom In or Zoom Out buttons are pressed. The calculator simply tells the model that it changed by X rate, then asks it for the scale, then passes it to AxesDrawerView which passes it to AxesDrawer, it draws, and boom, it's there.
There's only 1 problem which is actual drawing of the graph of points over pixels. Therefore I allocated AxesDrawerViewController at the initialization of CalculatorViewController, set its expression, and set its delegate.
The expression is what is recorded with each button pressed. For example I pressed 5 then + then X, so the expression will be 5 + X.
At the initialization of AxesDrawerViewController, I set the AxesDrawerView's delegate to be the AxesDrawerViewController. Therefore I can simply get the delegate through AxesDrawerView from the delegate by adding a method to AxesDrawerViewController which asks its delegate, which is CalculatorViewController to return the expression. Then I make an instance of CalculatorBrain, send it the expression, and the dictionary with the value of X I'm iterating over, so I can return it to AxesDrawer whenever it asks me to using a delegate.
The problem is that AxesDrawer's methods are class methods, so it can't see delegates. I don't know how to bypass this problem.
Any help is appreciated. Thanks.

Solaris 10 will not boot from DVD on my system

I first installed Solaris 10 on my system and it worked fine. Then when it went to boot the OS it just kept rebooting. I must have made a midtake in the install so went to install it again.
Problem is that now the DVD will not boot and give me the option to install Solaris like it did before. It instead goes straight into GRUB. I thought that this was because of the previous install on the hardrive. But then I disconnected my harddrives and still the same problem.
I tried the disk on another system and the disk booted with the option to install Solaris. So the question is why does the disc boot into GRUB even when there is no hard drives attached to my machine

as was mentioned, let's see /etc/vfstabHere it is:
#device to mount     #device to fsck     mount point     FS type     fsck pass    mount at boot       mount options
fs     -     /dev/fs     fs     -     no     -
/proc     -     /proc     proc     -     no     -
/dev/dsk/c0t0d0s1     -     -     swap     -     no     -
/dev/dsk/c0t0d0s0     /dev/rdsk/c0t0d0s0     /     ufs     1    no     -
/dev/dsk/c0t0d0s5     /dev/rdsk/c0t0d0s5    /usr    ufs    1    no     -
/dev/dsk/c0t0d0s4     /dev/rdsk/c0t0d0s4    /var/   ufs     1     no     -
/dev/dsk/c0t0d0s7     /dev/rdsk/c0t0d0s7    /export/home     ufs     2     yes     -
/dev/dsk/c0t0d0s3     /dev/rdsk/c0t0d0s3     /opt     ufs     2     yes     -
/devices     -    /devices     devfs     -    no    -
ctfs     -     /system/contract     ctfs     -     no     -
objfs     -     /system/object    objfs     -    no     -
swap     -     /tmp     tmpfs     -    yes     -
/dev/dsk/c0t1d0s0     /dev/rdsk/c0t1d0s0     /data     ufs     1     yes     -Sorry for the formatting, I had to type it out since I can't get any files directly off of the server.
The last entry in the vfstab file is the only entry I added myself. It is a secondary hard disk that I mounted to the directory /data. Everything else was automatically created at install according to the partitions that I defined.
I really need to get this server up and running as fast as possible. I want to avoid re-installing everything since that would be really time consuming. Thanks in advance.

FilesystemMountPoints for ufs disks mounted to non-global zones

Hello,
I have a SAN ufs disk to be used as a failover storage, mounted to non-global zones (NGZ).
Solaris 10 nodes using Cluster 3.2
I'm looking for the correct value for the property FilesystemMountPoints and the vfstab entry required for a failover disk mounted to a NGZ.
Should the path NOT include the NGZ root path?
From the man page for SUNW.HAStoragePlus, for the property FilesystemMountPoints:
You can specify both the path in a non-global zone and the path in a global zone, in this format:
Non-GlobalZonePath:GlobalZonePath
The global zone path is optional. If you do not specify a global zone path, Sun Cluster assumes that the path in
the non-global zone and in the global zone are the same. If you specify the path as
Non-GlobalZonePath:GlobalZonePath, you must specify Global-ZonePath in the global zone's /etc/vfstab.
The default setting for this property is an empty list.
You can use the SUNW.HAStoragePlus resource type to make a file system available to a non-global zone. To enable
the SUNW.HAStoragePlus resource type to do this, you must create a mount point in the global zone and in the
non-global zone. The SUNW.HAStoragePlus resource type makes the file system available to the non-global zone
by mounting the file system in the global zone. The resource type then performs a loopback mount in the
non-global zone.
Each file system mount point should have an equivalent entry in /etc/vfstab on all cluster nodes and in all
global zones. The SUNW.HAStoragePlus resource type does not check /etc/vfstab in non-global zones.
SUNW.HAStoragePlus resources that specify local file systems can only belong in a failover resource group
with affinity switchovers enabled. These local file systems can therefore be termed failover file systems. You
can specify both local and global file system mounts points at the same time.
Any file system whose mount point is present in the FilesystemMountPoints extension property is assumed to
be local if its /etc/vfstab entry satisfies both of the following conditions:
1. The non-global mount option is specified.
2. The "mount at boot" field for the entry is set to "no."
In my situation, I want to mount the disk to /mysql_data on the NGZ called ftp_zone. So, which is the correct setup?
a. FilesystemMountPoints=/mysql_data:/zones/ftp_zone/root/mysql_data
Global zone vfstab entry /dev/md/ftpabin/dsk/d110 /dev/md/ftpabin/rdsk/d110 /zones/ftp_zone/root/mysql_data ufs 1 no logging
NGZ mount point /mysql_data
OR
b. FilesystemMountPoints=/mysql_data:/mysql_data (can be condensed to simply /mysql_data)
Global zone vfstab entry /dev/md/ftpabin/dsk/d110 /dev/md/ftpabin/rdsk/d110 /mysql_data ufs 1 no logging
NGZ mount point /mysql_data
Should the path NOT include the NGZ root path?
And should the fsck pass # be 1 or 2?
Looking at this example from p. 26 of
http://wikis.sun.com/download/attachments/24543510/820-4690.pdf
This example doesn't mention the entry in vfstab.
Create a resource group that can holds services in nodea zonex and nodeb zoney
nodea# clresourcegroup create -n nodea:zonex,nodeb:zoney test-rg
Make sure the HAStoragePlus resource is registered
nodea# clresourcetype register SUNW.HAStoragePlus
Now add a UFS [or VxFS] fail-over file system: mount /bigspace1 to failover/export/install in NGZ
nodea# clresource create -t SUNW.HAStoragePlus -g test-rg \
-p FilesystemMountPoints=/fail-over/export/install:/bigspace1 \
ufs-hasp-rs
Thank you!

Hi,
/zones/oracle-z is my root directory of the zone.
* add the device to the zone :
root@mpbxapp1 # zonecfg -z oracle-z
zonecfg:oracle-z> add device
zonecfg:oracle-z:device> set match=/dev/global/dsk/d12s0
zonecfg:oracle-z:device> end
zonecfg:oracle-z> add device
zonecfg:oracle-z:device> set match=/dev/global/rdsk/d12s0
zonecfg:oracle-z:device> end
zonecfg:oracle-z> exit
* add FS to NGZ's /etc/vfstab : ( You may omit this step, I don't know why but it works without this step :) )
root@mpbxapp1 # vi /zones/oracle-z/root/etc/vfstab
/dev/global/dsk/d12s0 /dev/global/rdsk/d12s0 /global/oracle ufs 1 no logging
* add FS to global zone's /etc/vfstab :
root@mpbxapp1 # vi /etc/vfstab
/dev/global/dsk/d12s0 /dev/global/rdsk/d12s0 /zonefs/oracle ufs 1 no logging
* set the FilesystemMountPoints property :
root@mpbxapp1 # /usr/cluster/bin/clresource set -p FilesystemMountPoints=/global/oracle:/zonefs/oracle oracle-hastp
Whit this configuration you may ensure that the FS is not directly accessible from master zone. Actually, it's accessible but with a different PATH. For example, for Oracle, from the master zone Oracle can not be started/stopped because the controlfile can not be accessed. :)
Hope this helps,
Murat

[SOLVED] fsck pass

Similar Messages

Maybe you are looking for