Mails stuck on the server

[criptus]

Users (all on the local network) sometimes complain that mails they send somehow stuck on the server for too long. Sometimes there are even mails not delivered to the recipients! Some mails are marked as undelivered some would reach recipinet after 2-3 hours or even more, some, as they claim are simply lost! Incoming mails however work fine, also LAN and Internet access.

SX Server v. 11.4.3

What shoul I do?

[criptus]

After restarting server (linux actualy) stuck mails went out. However same problem soon appeared again!

I checked services. Item Structure Server was down and I started it. Problem remains.

Queues list shows only few old corrupted messages.

Please give me some clues! Users are despared and I'm getting too !

[mikethebike]

Hi,
where are the corrupted messages? Try removing them with omcontain (or omresub, with the delete option)
You may want to check the audit log to see if messages are actually being processed (tail -f ~scalix/logs/audit)

Also check the fatal log to see if there are any errors.

Mick

[criptus]

I'd expect 'pending' mails listed on queue list, but they aren't. Outlook set them as sent mail. Where are those mails waiting and why?

I'll try to do as suggested by Mike.

[mikethebike]

you may get a better idea of what the server is doing by usig "omqstat -t", at least you will see what the throughput is, and number of messages on each queue.

Mick

[criptus]

/var/log/messages says this:

Oct 3 15:41:45 sx smartd[5840]: Device: /dev/sdb, 33 Currently unreadable (pending) sectors
Oct 3 15:41:45 sx smartd[5840]: Device: /dev/sdb, 33 Offline uncorrectable sectors

Items structure server service remains down everytime I check services. Service event loger shows this:

REPORT Item Structure(Item Structure) 09.30.11 09:39:07
[OM.DMON 602] Item Structure Log Server stop

There is NO new queues not even error messages although 'stuck messages' problem still remains.

After restaring entire system at east once per day messages would reach their addresses... this is temporary solution at the moment.

I suspect /var/log/messages warning ????

Any ideas?

[mikethebike]

when you suspect mails are queued, what does omqstat tell you?
Are messages sent between users on the same server delivered within reasonable times (within seconds)?

Mick

[criptus]

There is NO omqstat command

[mikethebike]

~/diag/omqstat ?

maybe they removed it?

[ls-al]

~/diag/omqstat ?

maybe they removed it?

any chance you are referring to omqdump?
I couldnt find anything about a tool named omqstat even in the OTNs

Regarding other things discussed here:

The "Item structure server" is not started by default intentionally. Its unrelated to the mail delivery.

To check the mailflow: Configure audit logging at least for "Service router", "Internet Gateway" and "Local Delivery". A value of "11" gives most results. Logging goes to "/var/opt/scalix/??/s/logs/audit" (note that theres no logrotation by default, use sxmaint -daily).

Once you see that a routing entry to "UNIX:MIME" in audit, the message was given to sendmail and you should see them in the sendmail log and/or mailq. In most cases I have seen when a customer complains about a slow mail delivery and heavy queueing it was caused by DNS issues like deprecated entries in resolv.conf and comparable stuff.

cheers,
Dirk

[criptus]

I found cfg file (sys/audit.cfg), but have no idea how to set what suggested

[ls-al]

What is scalix audit conf file name?

please use the tool omconfaud to configure the logging.

Code: Select all

[root@test ~]# omshowaud
Service Router                                                   0
Local Delivery                                                   0
Internet Mail Gateway                                            0
Local Client Interface                                           0
Remote Client Interface                                          0
Administration                                                   0
Request Server                                                   0
Directory Synchronization                                        0
Bulletin Board Server                                            0
Background Search Service                                        0
POP3 interface                                                   0
Omscan Server                                                    0
Archiver                                                         0
[root@test ~]# omconfaud sr 11 ld 11 unix 11
omconfaud : Logging level updated OK.
[root@test ~]# omshowaud
Service Router                                                   11
Local Delivery                                                   11
Internet Mail Gateway                                            11
Local Client Interface                                           0
Remote Client Interface                                          0
Administration                                                   0
Request Server                                                   0
Directory Synchronization                                        0
Bulletin Board Server                                            0
Background Search Service                                        0
POP3 interface                                                   0
Omscan Server                                                    0
Archiver                                                         0


Theres also a man-page

[criptus]

Thnx, I'II report.

[criptus]

Once you see that a routing entry to "UNIX:MIME" in audit, the message was given to sendmail and you should see them in the sendmail log and/or mailq. In most cases I have seen when a customer complains about a slow mail delivery and heavy queueing it was caused by DNS issues like deprecated entries in resolv.conf and comparable stuff.

cheers,
Dirk

Where/how can I see mails stuck on the server? Queue list on sxadmin web console is empty although mails are waiting somewhere...

[ls-al]

SAC an command like omstat will only report messages that are still in Scalix and not delivered to local mailboxes or to sendmail. To see the queue of sendmail use "mailq".

Code: Select all

[root@test ~]# mailq
/var/spool/mqueue is empty
      Total requests: 0