Cleaning /tmp on a SunRay server

Question

Hello,
I'm running a server with 70 SunRay units, used by a different group of users every week. Mostly. Some stick around for a few months, but the majority are here for a week at a time. The /tmp directory gets littered quickly with files related to SunRay usage. We're also using Sun's GNOME 2.0 desktop environment, which creates /tmp files. Some of the files in /tmp should not be deleted, since they will cause the SunRay units to go into a "transient state", where they have an IP address, but no graphics data is arriving from the server. The screen will display only a blue box containing the IP and MAC addresses, a status code, etc, and it will float around the screen like a screensaver. The same thing appears to happen after about 30 days, to random SunRay units, if no files from /tmp are deleted. I am trying to determine which files, if deleted, will cause the floating blue box of death, and which ones I should nuke to maintain stability.
Here are some examples of files from /tmp that I'm not sure when or if to erase:
-rw------- 1 root other 352 Aug 30 15:07 .Xauth.Aay8j
drwxrwxr-x 2 root root 1536 Sep 28 14:28 .X11-pipe
-rw-rw-rw- 1 root other 0 Aug 10 18:11 X101
prw-rw-rw- 1 root other 0 Sep 28 08:25 X11
prw-rw-rw- 1 root other 0 Sep 28 08:39 X12
-rw-rw-rw- 1 root other 0 Sep 27 20:39 X13
drwxrwxr-x 2 root root 1536 Sep 28 14:28 .X11-unix
srwxrwxrwx 1 root other 0 Sep 28 01:25 X10
srwxrwxrwx 1 root other 0 Aug 10 18:11 X100
srwxrwxrwx 1 root other 0 Aug 10 18:11 X101
What is a good way to tell if the .Xauth* files, and the files in the .X11-unix and .X11-pipe dirs, are still in use or are from a now non-existent X session?
-rw-rw-rw- 1 root other 1388 Sep 28 01:26 .dcs.robson:10.37dd79
srwxr-xr-x 1 jbullock 03frg304 0 Sep 23 07:03 jpsock.140_03.12710
-rw------- 1 jbullock 03frg304 49 Sep 23 06:58 sh12690.2
There are dozens of .dcs* and sh* files. What are the .dcs*, jpsock.* and sh* files for, and when is it safe to nuke them?
Many thanks,
Brent

Answer 1

i had the "Mailbox is on a different server" error too recently (mine shows up in ms outlook express) - and believe my problem came from having the hostnamealiases space delimited instead of comma separated (with no spaces, e.g. "foo.host1.com,bar.host2.com") - comma delimited, no spaces seems to be required ...
i am having a different problem now though:
i can login to communications express hosted domains via -
the main host url using credentials like: [email protected]/passwd
but when going to any of the hosted.com's urls, and trying to login without the @hosted1.com in the user - i get redirected to access manager (and subsequently can't login using either uid or [email protected] at the hosted1.com/uwc url).
here's what i have in AMConfig.properties:
com.sun.identity.server.fqdnMap[mail.hosted1.com]=mail.actual.org
com.sun.identity.server.fqdnMap[mail.hosted2.com]=mail.actual.org
com.sun.identity.server.fqdnMap[mail.hosted3.com]=mail.actual.org
here's what i have in uwcauth.properties:
virtualdomain.mode = y
mail.actual.org.isvirutalhostname=mail.actual.org
i have also added the cookiedomains in amconsole under service configuration --> platform - in addition to creating directories (hosted1.com, hosted2.com, hosted3.com) under /opt/SUNWuwc/WEB-INF/domain - containing the original files and directories under /opt/SUNWuwc/WEB-INF/domain (i was not clear if these directories should take this form e.g. including the .com or if they should just contain the domain-name --e.g. hosted1?)
i can receive hosted email via pop using uid [email protected]/passwd, etc.
thanks for any input.
s7
using sparc 2005q4 msg -58; uwc - 42