CreateFile() taking a long time for small files on disk (sometimes)

Ever since upgrading a server of mine from Server 2008 R2 to Server 2012, I've been having trouble with shared files not serving quickly at times. I've traced the problem all the way back to the local machine. CreateFile() can take up to a minute
to return for small (few KB) files on disk, but it doesn't happen all the time. It happens more when the file is being requested by a lot of clients via network shares.
Here is an example. The top shows an example of a long request while the bottom shows a normal request.
I'm curious if anyone might know the cause for this sort of issue. I would think read cache alone would prevent this sort of thing from happening, and even if it had to come from disk each time I don't understand how it could take so long. The
disks are basically unloaded during these times.
This problem only seems to occur for files being served over the network. If I start randomly loading a copy of these files with even 128 different threads, I don't get any sort of weird contention issues if the file is local and not shared. Maximum
service times peak at 0.009s. SMB must be blocking the files somehow? As soon as the file is shared across the network, I start having the weird blocking issue even for local file requests on the server when the requests are to the local path of
those shared files.

No change in behavior with Server 2012 R2 and years of patches since I first posted.
I recently finished building and configuring a new storage server based on Server 2012 R2 Standard. Initially I thought that whatever was wrong must have been fixed in 2012 R2 since performance was good all week.
Over the weekend I moved the one remaining share from the old 2012 server to the new 2012 R2 server. The old server is the one that had had SMB2/3 disabled. The new server I had just left configured as default with SMB2/3 enabled and hadn't had
any trouble. This is the share I originally posted about in 2013 which has around 1.2 million small text files.
Starting this morning as employees started coming in and working, problems were immediately noticed. Slow performance was back and with a vengeance. I've disabled SMB2/3 on the 2012 R2 server now. It's disappointing that an issue like this
can go unnoticed and unpatched by Microsoft for so long. I don't have an unusual usage scenario. I have a mix of Windows 2000, XP, 7, and 8.1 clients that access the file server. Only XP, 7, and 8.1 clients access the problematic share.
I noticed this time around that the event log is flooded with SMBServer 1006 events saying that authentication failed for users accessing shares. I guess the clients auto-retry in a transparent fashion since clients typically don't have an issue other
than random-length delays--outright failures are rare. The shares gives the specified users full access as do the disk permissions for the share's folder. There isn't any logical reason for an authentication failure other than some systemic flaw.
Unfortunately, I don't have any answers. I wanted to keep this thread up to date however for other users in the future to reference.

Similar Messages

Taking too long time for booting

Hi there,
I just bought myself an iphone 5s. . Try to get up started for the 1st time but the thing keep booting or loading too long time for endless hours. What may go wrong? Appreciate if you could help me on this.

But payment run before due date. you are not run before due date once check it

My mac pro is taking a long time to open files

When I rey and open applications like Finder, it is taking a long time to open or it hangs how can I fix this
I have OS10.7.2 and 300GB left out of 500GB.
Mac book peo 17
Message was edited by: waldoamt

How large is your hard drive and how much hard drive space do you have left?
Which Macbook Pro do you have?
Which os & version are you using?

Why iPad2 is taking too long time for a software update?

Hi, i have iPad2 with iOS 4.3, now I'd like to update to iOS 6.1.2, but it is taking too long time to update when i connect to iYunes.My internet speed is 15mbps.I am unable to understand the problem.Please help me

Hi, i have iPad2 with iOS 4.3, now I'd like to update to iOS 6.1.2, but it is taking too long time to update when i connect to iYunes.My internet speed is 15mbps.I am unable to understand the problem.Please help me

Mail taking a long time to preview files to be attached

I am wondering if anybody can help. Mail is taking ages to preview a file (large or small) before they can be attached. I start seeing the beach ball of death spinning for a long while (> 1 min) before it eventually shows a preview of the file highlighted. I first noticed this problem with finder, where it was taking ages to preview a file...i found this link: http://www.thexlab.com/faqs/finder.html
and it works finder is fine after following those steps...But i tried the same thing with mail, i.e. delete the com.apple.mail.plist file, and mail starts all over (have to be set up again) but it still is giving this problem. And once i try this in mail, get this problem, it automatically leads to the same problem again with finder. Should i also say i wasn't having this problem until the new secruity update? Can anyone help?
Thanks,
Kurt

I suggest another link provided by Dr. Smoke - Resolving Disk, Permission, and Cache Corruption for system troubleshooting procedures and when doing so, follow all steps/instructions in the order provided.

Batch creation program takes long time for large file

Hi,
I am uploading batch using a custom program which uses BAPI_BATCH_SAVE_REPLICA.The program takes 4 hrs for uploading 100000 records.
But when I am using a file with 400000 records. It creates the second batch after 8 hrs from the starting of the batch job. The job takes
more than 50 hrs to complete.
Any suggestion?
Regards,
Tan

We had a simular problem loading materials where it would have taken over 23 hours to lad over 500,000 material masters. It may be that Logged changes are switched on to the table that you are loading the data into and that te,porairy you need to get this switched off or copy and amend the bapi that you are using so that it doesn't update the log.
The other thing that you can do is to split your data file into smaller batches and load it that way.

Middleware- Taking long time for generation of Runtime objects- SMOGTOTAL

Hi Experts,
I am doing middleware settings for connecting CRM 2007 with R/3 4.7.
When i am generating all the required objects ( Replication objects, publications....) using the transaction code SMOGTOTAL, system is taking very long time for generating the objects. Generally it takes 4 to 6 hours but in our case it has already took more than 36 hours and still its running.
Can anybody tell me what i need to do to make the generation process faster.
Regards
Nadh

What I read in the best practice:
It is not required for a new installation.
Typically this activity has already been executed during the system installation or upgrade.
Use transaction SMOGLASTLOG to check if an initial generation has already been executed. In this case you can skip this activity.
I checked transaction SMOGLASTLOG, and in our case the initial generation was not yet executed. I also couldn't continue with the next steps.
That's why I started up the job, it is finally finished after 104 hours..
Thanks for your fast reply.
Jasper.

Longer time for Material Availability check while creation of prd order.

Hi guys,
I am facing a weird problem while creating production orders thru CO01.
I enter the component and plant and I am also using the forward scheduling option.
for some reason, SAP is taking a long time for material availability check when I hit the release button.
Sometimes its taking more than an hour. Its happening with few specific BOM's, and I have checked the master data but I could hardly find a problem in master data.
Can someone suggest me some tips ??
Thanks & Regards,
Sashivardhan

Hi,
Please check the Availability check control maintained for Components it should be 01 or 02. Also check the issue storage location maintained or not. You can maintain issue storage location in BOM in Status/lng text tab in Production Storage Location.
Hope this will help.
Regards,
Navin

DIAdem taking a long time to load 700MB TDMS file to data portal

I'm trying to load a 700MB TDMS file into DIAdem ,4 channels 10Hz sampling rate over 4 days . When i drag the file into the data portal , diadem freezes up , its on a 2GB machine, file located on the desktop, not running any other programs and it went 50min without being able to load before i made it quit. seems to work fine on smaller files though . Was wondering if this sounds normal and if anyone knows a way around it.
yin
Solved!
Go to Solution.

Thanks for all your input.
YongqinYe and Brad : I've tried defragmenting but this also seems to be taking a long time (in excess of ) on 2 different machines. The index file is 300MB, so about 75% of the *.tdms,
Otmar: i also tried registry loading the data but this still took a long time on this particular file although it worked great on smaller files (5MB).
Brad : the file does not open in the labVIEW TDMS viewer although i'm able to view it in chunks of 65000 using the TDMS importer for excel.
Stefan: the file does have explicit time stamp channel .
Any extra help would be appreciated but the way I'm thinking of doing it now is to use much longer buffer size (from only 6 to a few hundred) when collecting the data to decrease the level of fragmentation.

Query taking long time for EXTRACTING the data more than 24 hours

Hi ,
Query taking long time for EXTRACTING the data more than 24 hours please find the query and explain plan details below even indexes avilable on table's goe's to FULL TABLE SCAN. please suggest me.......
SQL> explain plan for select a.account_id,round(a.account_balance,2) account_balance,
2 nvl(ah.invoice_id,ah.adjustment_id) transaction_id,
to_char(ah.effective_start_date,'DD-MON-YYYY') transaction_date,
to_char(nvl(i.payment_due_date,
to_date('30-12-9999','dd-mm-yyyy')),'DD-MON-YYYY')
due_date, ah.current_balance-ah.previous_balance amount,
decode(ah.invoice_id,null,'A','I') transaction_type
3 4 5 6 7 8 from account a,account_history ah,invoice i_+
where a.account_id=ah.account_id
and a.account_type_id=1000002
and round(a.account_balance,2) > 0
and (ah.invoice_id is not null or ah.adjustment_id is not null)
and ah.CURRENT_BALANCE > ah.previous_balance
and ah.invoice_id=i.invoice_id(+)
AND a.account_balance > 0
order by a.account_id,ah.effective_start_date desc; 9 10 11 12 13 14 15 16
Explained.
SQL> select * from table(dbms_xplan.display);
PLAN_TABLE_OUTPUT
| Id | Operation | Name | Rows | Bytes |TempSpc| Cost (%CPU)|
| 0 | SELECT STATEMENT | | 544K| 30M| | 693K (20)|
| 1 | SORT ORDER BY | | 544K| 30M| 75M| 693K (20)|
|* 2 | HASH JOIN | | 544K| 30M| | 689K (20)|
|* 3 | TABLE ACCESS FULL | ACCOUNT | 20080 | 294K| | 6220 (18)|
|* 4 | HASH JOIN OUTER | | 131M| 5532M| 5155M| 678K (20)|
|* 5 | TABLE ACCESS FULL| ACCOUNT_HISTORY | 131M| 3646M| | 197K (25)|
| 6 | TABLE ACCESS FULL| INVOICE | 262M| 3758M| | 306K (18)|
Predicate Information (identified by operation id):
2 - access("A"."ACCOUNT_ID"="AH"."ACCOUNT_ID")
3 - filter("A"."ACCOUNT_TYPE_ID"=1000002 AND "A"."ACCOUNT_BALANCE">0 AND
ROUND("A"."ACCOUNT_BALANCE",2)>0)
4 - access("AH"."INVOICE_ID"="I"."INVOICE_ID"(+))
5 - filter("AH"."CURRENT_BALANCE">"AH"."PREVIOUS_BALANCE" AND ("AH"."INVOICE_ID"
IS NOT NULL OR "AH"."ADJUSTMENT_ID" IS NOT NULL))
22 rows selected.
Index Details:+_
SQL> select INDEX_OWNER,INDEX_NAME,COLUMN_NAME,TABLE_NAME from dba_ind_columns where
2 table_name in ('INVOICE','ACCOUNT','ACCOUNT_HISTORY') order by 4;
INDEX_OWNER INDEX_NAME COLUMN_NAME TABLE_NAME
OPS$SVM_SRV4 P_ACCOUNT ACCOUNT_ID ACCOUNT
OPS$SVM_SRV4 U_ACCOUNT_NAME ACCOUNT_NAME ACCOUNT
OPS$SVM_SRV4 U_ACCOUNT CUSTOMER_NODE_ID ACCOUNT
OPS$SVM_SRV4 U_ACCOUNT ACCOUNT_TYPE_ID ACCOUNT
OPS$SVM_SRV4 I_ACCOUNT_ACCOUNT_TYPE ACCOUNT_TYPE_ID ACCOUNT
OPS$SVM_SRV4 I_ACCOUNT_INVOICE INVOICE_ID ACCOUNT
OPS$SVM_SRV4 I_ACCOUNT_PREVIOUS_INVOICE PREVIOUS_INVOICE_ID ACCOUNT
OPS$SVM_SRV4 U_ACCOUNT_NAME_ID ACCOUNT_NAME ACCOUNT
OPS$SVM_SRV4 U_ACCOUNT_NAME_ID ACCOUNT_ID ACCOUNT
OPS$SVM_SRV4 I_LAST_MODIFIED_ACCOUNT LAST_MODIFIED ACCOUNT
OPS$SVM_SRV4 I_ACCOUNT_INVOICE_ACCOUNT INVOICE_ACCOUNT_ID ACCOUNT
OPS$SVM_SRV4 I_ACCOUNT_HISTORY_ACCOUNT ACCOUNT_ID ACCOUNT_HISTORY
OPS$SVM_SRV4 I_ACCOUNT_HISTORY_ACCOUNT SEQNR ACCOUNT_HISTORY
OPS$SVM_SRV4 I_ACCOUNT_HISTORY_INVOICE INVOICE_ID ACCOUNT_HISTORY
OPS$SVM_SRV4 I_ACCOUNT_HISTORY_ADINV INVOICE_ID ACCOUNT_HISTORY
OPS$SVM_SRV4 I_ACCOUNT_HISTORY_CIA CURRENT_BALANCE ACCOUNT_HISTORY
OPS$SVM_SRV4 I_ACCOUNT_HISTORY_CIA INVOICE_ID ACCOUNT_HISTORY
OPS$SVM_SRV4 I_ACCOUNT_HISTORY_CIA ADJUSTMENT_ID ACCOUNT_HISTORY
OPS$SVM_SRV4 I_ACCOUNT_HISTORY_CIA ACCOUNT_ID ACCOUNT_HISTORY
OPS$SVM_SRV4 I_ACCOUNT_HISTORY_LMOD LAST_MODIFIED ACCOUNT_HISTORY
OPS$SVM_SRV4 I_ACCOUNT_HISTORY_ADINV ADJUSTMENT_ID ACCOUNT_HISTORY
OPS$SVM_SRV4 I_ACCOUNT_HISTORY_PAYMENT PAYMENT_ID ACCOUNT_HISTORY
OPS$SVM_SRV4 I_ACCOUNT_HISTORY_ADJUSTMENT ADJUSTMENT_ID ACCOUNT_HISTORY
OPS$SVM_SRV4 I_ACCOUNT_HISTORY_APPLIED_DT APPLIED_DATE ACCOUNT_HISTORY
OPS$SVM_SRV4 P_INVOICE INVOICE_ID INVOICE
OPS$SVM_SRV4 U_INVOICE CUSTOMER_INVOICE_STR INVOICE
OPS$SVM_SRV4 I_LAST_MODIFIED_INVOICE LAST_MODIFIED INVOICE
OPS$SVM_SRV4 U_INVOICE_ACCOUNT ACCOUNT_ID INVOICE
OPS$SVM_SRV4 U_INVOICE_ACCOUNT BILL_RUN_ID INVOICE
OPS$SVM_SRV4 I_INVOICE_BILL_RUN BILL_RUN_ID INVOICE
OPS$SVM_SRV4 I_INVOICE_INVOICE_TYPE INVOICE_TYPE_ID INVOICE
OPS$SVM_SRV4 I_INVOICE_CUSTOMER_NODE CUSTOMER_NODE_ID INVOICE
32 rows selected.
Regards,
Bathula
Oracle-DBA

I have some suggestions. But first, you realize that you have some redundant indexes, right? You have an index on account(account_name) and also account(account_name, account_id), and also account_history(invoice_id) and account_history(invoice_id, adjustment_id). No matter, I will suggest some new composite indexes.
Also, you do not need two lines for these conditions:
and round(a.account_balance, 2) > 0
AND a.account_balance > 0
You can just use: and a.account_balance >= 0.005
So the formatted query isselect a.account_id,
       round(a.account_balance, 2) account_balance,
       nvl(ah.invoice_id, ah.adjustment_id) transaction_id,
       to_char(ah.effective_start_date, 'DD-MON-YYYY') transaction_date,
       to_char(nvl(i.payment_due_date, to_date('30-12-9999', 'dd-mm-yyyy')),
               'DD-MON-YYYY') due_date,
       ah.current_balance - ah.previous_balance amount,
       decode(ah.invoice_id, null, 'A', 'I') transaction_type
from account a, account_history ah, invoice i
where a.account_id = ah.account_id
   and a.account_type_id = 1000002
   and (ah.invoice_id is not null or ah.adjustment_id is not null)
   and ah.CURRENT_BALANCE > ah.previous_balance
   and ah.invoice_id = i.invoice_id(+)
   AND a.account_balance >= .005
order by a.account_id, ah.effective_start_date desc;You will probably want to select:
1. From ACCOUNT first (your smaller table), for which you supply a literal on account_type_id. That should limit the accounts retrieved from ACCOUNT_HISTORY
2. From ACCOUNT_HISTORY. We want to limit the records as much as possible on this table because of the outer join.
3. INVOICE we want to access last because it seems to be least restricted, it is the biggest, and it has the outer join condition so it will manufacture rows to match as many rows as come back from account_history.
Try the query above after creating the following composite indexes. The order of the columns is important:create index account_composite_i on account(account_type_id, account_balance, account_id);
create index acct_history_comp_i on account_history(account_id, invoice_id, adjustment_id, current_balance, previous_balance, effective_start_date);
create index invoice_composite_i on invoice(invoice_id, payment_due_date);All the columns used in the where clause will be indexed, in a logical order suited to the needs of the query. Plus each selected column is indexed as well so that we should not need to touch the tables at all to satisfy the query.
Try the query after creating these indexes.
A final suggestion is to try larger sort and hash area sizes and a manual workarea policy.alter session set workarea_size_policy = manual;
alter session set sort_area_size = 2147483647;
alter session set hash_area_size = 2147483647;

Taking very long time to pick the file from internal deliviery channel fold

Hi B2B gurus,
For one of our trading partner we are testing an AS2 connectivity, so we are sending a 850 po's through internal deliviery channel(create a folder in-house), but while placing the 850 in a internal folder, its taking very long time to pick the file form internal folder
Could you please help me to resolve this issue?
regards

Hi Suni,
Have you specified the polling interval in Transport Server configuration? By default polling interval is 60 seconds.
What is the size of the file being picked?
Regards,
Anuj

Impdp taking long time for only few MBs data...

Hi All,
I have one query related to impdp. I have one expdp file and size is 47M. When I restore this dmp using impdp it will take more time. Also initially table_data loaded finsih very fast but then later on alter function/procedure/view taking a lot time almost 4 to 5 hrs.
I have no idea why its taking long time... Earlier I can see one DB link is failed and it given error for "TNS name could not resolved" so I create DB link also before run impdp but the same result. Can any one suggest what could be the cause to taking long time for only 47MB data?
Note - Both expdp and impdp database version is 11.2.0.3.0. If I import the same dmp file into 11.2.0.1.0 then its done in few mins.
Thanks...

Also Read
Checklist For Slow Performance Of DataPump Export (expdp) And Import (impdp) [ID 453895.1]
DataPump Import (IMPDP) is Very Slow at Object/System/Role Grants, Default Roles [ID 1267951.1]

Why is it taking a very long time to migrate files files macbook to iMac?

It is taking a very long time to migrate files from a macbook to a new iMac (24 hours). Is this normal for around 40GB of data?

Here is some info on various methods:
http://support.apple.com/kb/HT6025
And if you're using Mavericks:
http://support.apple.com/kb/HT5872
And more good info:
http://pondini.org/OSX/Setup.html

I just downloaded the new update for itunes and the itunes store wont open up it is taking a long time it still hasnt loaded since 5pm

i just downloaded the new update for itunes and the itunes store wont open up it is taking a long time it still hasnt loaded since 5pm, i also cant sighn in either it just keeps on saying accessing itunes....

Close your iTunes,
Go to command Prompt -
(Win 7/Vista) - START/ALL PROGRAMS/ACCESSORIES, right mouse click "Command Prompt", choose "Run as Administrator".
(Win XP SP2 & above) - START/ALL PROGRAMS/ACCESSORIES/Command Prompt
In the "Command Prompt" screen, type in
netsh winsock reset
Hit "ENTER" key
Restart your computer.
If you do get a prompt after restart windows to remap LSP, just click NO.
Now launch your iTunes and see if it is working now.
If you are still having these type of problems after trying the winsock reset, refer to this article to identify which software in your system is inserting LSP:
iTunes 10.5 for Windows: May see performance issues and blank iTunes Store
http://support.apple.com/kb/TS4123?viewlocale=en_US

Every 3rd data package taking long time for execution

Hi Everyone
We are facing a strange situation. Our scenario involves doing a full load from DSO to CUBE.
Start routines are not very database intensive and care has been taken to write them in a optimized way.
But strangely every 3rd data package is taking exceptionally longer time than other data packages.
a) DTP is having 3 parallal processes.
b)time spent in extraction , rule, and updation is constant for every data package.
c)start routine time is larger for every 3rd data package and keeps on increasing. for e.g. 5 mins, 10 mins, 24 mins, 33 mins etc it increases by each 3rd package.
I tried to anlayze the data which was taking so much time but found no difference in terms of data in normal and longer time taking DTP (i.e. there was not logical difference in data for start routine to behave like this).
I was wondering what can be the possible reasons for it and may be some other external system factors can be responsible for it. If someone can help in this regard that will be highly appreciated.

Hi Hemanth,
In your start routine, are you by any chance adding or multiplying the number of records to the source_package? Something like copy source package into an internal table, add records to internal table and then copy it back to source package? If some logic of this sorts is in your start routine, you need to refresh your internal table. Otherwise, the internal table records goes on increasing with every data package. So, the processing time might increase as the load progresses. This is one common mistake I have seen. Please check your code if you have something like that and refresh the internal tables. See if this makes any difference.
Thanks and Regards
Subray Hegde

CreateFile() taking a long time for small files on disk (sometimes)

Similar Messages

Maybe you are looking for