sxmkindex fails for some users

[Kris]

Yesterday I installed a Scalix 11.1 server, to test the upgrade from our current 10.0.5 server.

Now I want to run sxmkindex to create the index for all users. For most of the users it works fine, but for some users it gives the message: user /mailnode: oops: No such file or directory

I tried disabeling indexing from SAC, then enabling it again and then recreate the index, but this gives the same error again. This happens for about 15-20 mailboxes on a total of about 225.

Any suggestions?

[jch]

That sounds as though there's something badly wrong with the account. Does "omscan -a" report anything amiss? Can you do an active scan for the bad user? eg

Code: Select all

omscan -AvxU user/mailnode

Does that show any problems?

jch

[Kris]

Thanks for your reply.

I just did an omscan -AvxU for one of the users with this problem:

Code: Select all

[root@scalixserver /]# omscan -AvxU "<user>"/<mailnode>
omscan running on 09.26.07 at 16:10:36.
Host computer : scalixserver.domain.com.
Report mode requested.

Last omscan tool run on 09.26.07 at 16:10:15; duration 1 minute(s).
Previous server cycle run on 09.19.07 at 07:48:24; duration 221 minute(s).
Current server cycle started on 09.26.07 at 07:48:24; in progress.

Extra check option requested.
Active scan option requested.

Scanning file/dir links .... done.

CAUTION: Scanning of message store has started.
         Mounted file/dir links must be maintained during the scan.
         VxFS file system must not be reorganized - see omscan(1M).

Checking/Scanning user trays ....<user> / ScalixServer, nido/CN=<user>
 done.

Missing children ....

Owner Info : <user> / ScalixServer, nido/CN=<user>
Parent Container : ~/user/g00004r/0000001:1, RecNum : 4681
Missing Child : ~/data/000004n/00h1ose:2
Child Type : Container.
Child Creator : crescentsuteb /internet (RFC-822|crescentsuteb@taele-party-c)
Child Subject : Ready, right now

Disk usage ....

USER NAME                          IN    OUT    PDG   FCAB   DLST     WB   TOTAL (KB)

<user> /<mailnode> 1319410      2 351193 4471341      1      1 6141948


I'm running an omscan -a now, but that takes a while..

[jch]

That's the problem then. There's a message in that user's INBOX that's broken. (I think a file called "0000001" is the INBOX anyway).

The message is missing and all that's left it a pointer to it (so we know who sent it and what the subject was) you can fix the problem by deleting the pointer -- just add "f" (for fix) to the beginning of the set of flags that you gave omscan.

You'll probably find similar problems on the other users with problems and if you do you can clear them up in the same way. (But always check what you're going to fix before you repair it, you never know what might have happened.)

jch

[Kris]

I ran the omscan with the -f option now to fix the problem. After that, I did a check again without the -f option. Now it finds a new problem.

Bad extensions ....

Owner Info : <user> / <mailnode>/CN=>user>
Parent Container : ~/user/g00004r/0000001:1, RecNum : 91
Child Affected : ~/data/00000cq/003blpm:1
Child Subject : <subject>

Now running omscan with the -f option again, see if this problem gets fixed now.. I'm curious why he didn't report this problem the first time..

[Kris]

Okay, also fixed this problem. Now, omscan doesn't report any more errors for this user.

I think sxmkindex will run okay now for this user, it's running now, and before it quit after just running for a very short time.

Thanks for your help!!

edit: yes, it's working now 8)