mail account corrupt?

[dijichi2]

i've imported about 15 mailboxes on average 500mb each to zimbra using imapsync. apart from imapsync barfing on messages over 10mb everything went fine, flags, tags and dates came in ok. but i've got one account thats sick - when i hit reindex in the admin interface using ie or firefox, the browser hangs solid and you get two java processes use a lot of cpu and memory for an extended period. if you leave the browser it eventually comes back saying resource could not be found for the reindex. browser also hangs with animated bar waiting to log into account.

is the a command line thingee for checking/reindexing an account? all others seem to work fine.

[tim brennan]

there is no command line reindex call, but you can take the perl script in ZimbraServer/src/perl/soap/reindex.pl and change the hardcoded URL and it should work -- it just sends the <reindexrequest> Soap command to the server.

I'm not sure what a CLI really gives you though -- I seriously doubt that a CLI will work if the web UI doesn't. What is wrong with the mailbox that you decided to reindex it? Can that user log on? Does new mail delivery work for that mailbox?

[tim brennan]

Ahh, oops, I missed the part where you said "browser also hangs with animated bar waiting to log into account."

Do you know what the mailbox ID is? Can you look in the index directory /opt/zimbra/index/0/[mailbox-id]/index/0 and post an ls of the directory?

Oh yah, also, the CPU hang sounds like a firefox bug (FF spins at 100% CPU while waiting for server to respond) that was fixed in FF 1.5...are you using an older version of FF?

[tim brennan]

You can find the mailbox ID using this two-step process:

First, get the AccountId of the user from LDAP:

% /opt/zimbra/bin/zmprov ga user@domain.com| grep -i zimbraId
zimbraId: LONG-ALPHANUMERIC-UID-STRING

Then use that AccountId to lookup their mailbox ID in MySQL:

% /opt/zimbra/bin/mysql -e "use zimbra; select id from mailbox where account_id = 'UID-STRING-FROM-ABOVE'"
+----+
| id |
+----+
| 26 |
+----+

[jesse]

I'm stuck in this same problem. How can I get this account fixed?

[zimbra@mail 0]$ pwd
/opt/zimbra/index/0/11/index/0
[zimbra@mail 0]$ ls
_8l.cfs  _bg.cfs  _eb.cfs  deletable  segments
[zimbra@mail 0]$

[konrad]

Same problem here, reindex on a mailbox stuck in the middle, now the account can't login anymore nor receive email (postfix shows timed out while sending end of data).

Running ZimbraServer/src/perl/soap/reindex.pl (action status, start, cancel) just hangs, running it on another mailbox does work.

Nice to see that there might be a bug filled for this. ( Index Interrupted - User Can't Login ) but this bug is not accessible by me ( You are not authorized to access bug #7725. )

I'm not running Network Edition. When I ran the reindex on the mailbox it was version 3.1.3 now upgraded to 3.1.4 but no way to fix the account.

[konrad]

tim it would be nice if you could get us any hints on wether these indexing errors can be recovered

[konrad]

Well for those people that would like to know how to recover from this here is a simple step by step guide (this was using 3.1.4):

1. Set the problematic account into maintance mode
2. Get the mailbox id of the account as described by Tim
3. Shutdown Zimbra
4. Delete the folder called /opt/zimbra/index/0/<MAILBOX ID>
5. Startup Zimbra - You will see that zimbra rebuilds the index of the mailbox.
6. Wait until the reindexing is finished
7. Make the account active again.

And here we are again, mailbox can receive email and user can login...

[tim brennan]

I'm not running Network Edition. When I ran the reindex on the mailbox it was version 3.1.3 now upgraded to 3.1.4 but no way to fix the account.

Yah, this was unfortunately a result of my initial 'fix' of 7725 (oops) -- while trying to fix one problem I inadvertantly created a deadlock that could be triggered during reindexing. You should be able to re-index again in 3.1.4 now.

If your index is really corrupt (seems like it might be) you will have to manually delete it before you can re-index. Konrad's instructions are good with the minor exception that the index might be in /opt/zimbra/index/1/<MAILBOX ID> (or /2/<MAILBOX_ID>) in some cases.

No handle on what might be causing the corruptions right now: we've seen very few (the ones in this thread are the only ones to my knowlege) instances of this happening -- so it is difficult to pin down.

[tim brennan]

A word of suggestion: you really should use the admin console's "reindex mailbox" button to trigger reindexing, not the perl script. The big advantage is that the admin console functionality gets run by the Test team, whereas the perl script does not....