Sophos UTM 9.509-3 Home Edition as HA - Hourly messages: Pop3 proxy not running, ACC device Agent not running, HA selfcheck

If you search here, you will see that this problem with VMs in HA first appeared in 9.508.  I don't think it was fixed in 9.509.  I think you will find that others solved the problem by going back to 9.506.

Cheers - Bob

Hi BAlfson,

sorry, I didn't find similar posts in the forum. And I didn't had this issue with 9.508. And with 9.509 there was no issue from March, 28 until yesterday.

Can you give me a link to another post with this issue, please? I want to see if this is really the same issue. Thank you.

Hallo,

I don't know that this is the same problem.  I just remember that there were problems with VMs in HA in 9.508.  I would try a Google:

site:community.sophos.com/products/unified-threat-management/f "9.508" VM

Cheers - Bob

Hi BAlfson,

these posts are about HTTP proxy. I have the issue with Pop3 proxy.

Today, I switched the preferred master to node 2. I want to see if this helps. An interesting thing is that the CPU load (95 %) on node 1 is still very high even if it's not the master anymore. Normally, the CPU load averages at up to 5 % on the master in my environment. On node 2 the CPU is running at normal load level. After restarting node 1 the CPU load first came back to normal load level but after a successful HA synchronization the CPU load increased to 45 %. Waiting some time it came back again to normal CPU load. 

After switching the master to node 2 Up2Date is working again. The patterns were updated and I got a success message per mail as configured. And node 1, now slave, got the updates after the restart, too.

Now, I want to see if there are hourly messages about "Pop3 proxy not running", "ACC device agent not running" and "HA selfcheck", today.

The daily executive report isn't working. When starting manually there's a little popup window with "Please be patient, while the report is generated." But the picture isn't shown and this window won't close automatically. And I don't get the report.

UPDATE, April, 7, 2018, 10:05 CEST: I still get messages regarding "Pop3 proxy not running" from both nodes and "HA selfcheck" in a hourly schedule. Only the ACC messages are gone. The CPU load of both nodes is normal.

UPDATE, April, 8, 2018, 06:55 CEST: There are still messages regarding "Pop3 proxy not running" from both nodes and "HA selfcheck" from the master node in a hourly schedule. I still don't get the daily executive report. The CPU load is at a normal level. I will now stop the slave node for today and will have a look, if this helps to prevent from the issue.

UPDATE, April, 8, 2018, 20:05 CEST: There are still hourly messages regarding "Pop3 proxy not running". And of course there are messages that the HA group is broken.

UPDATE, April, 9, 2018, 06:20 CEST: There are still hourly messages regarding "Pop3 proxy not running". And the daily executive report isn't working. I start node 1 to have HA running again.

Hi all,

after trying some things this weekend (see my post before with the updates) I don't know if this issue is related to a bug of Sophos UTM 9.509 or if there's an issue with the nodes itself. I updated to 9.509 early after the announcement and at first I didn't have errors or problems. And even if there are these messages, everything seems to work except the daily executive report.

Are there more ideas what to check? Thank you.

Hi all,

does no one else have an issue like me?

I'm thinking about reinstalling my both nodes with the iso file of Sophos UTM 9.509-3 and restoring my last good backup from the same version. Maybe this will help to not to get these messages anymore. And hopefully I will then get the daily executive reports again. Or has someone else other ideas how to solve the issues?

Will the Postgres database then be 64 bit? In the past there was a way to update the database from 32 to 64 bit manually on the console. But this isn't needed anymore, right?

Hi all,

tonight I installed a fresh HA pair of Sophos UTM with the ISO file of 9.509-3 and the restore of a backup, where there was no issue. And for the moment it looks like the issue is resolved. There are no more error messages. Now it's possible to create a daily executive report manually and to get this report as scheduled task again.

Regarding my question about the Postgres database: The web site https://community.sophos.com/kb/en-us/126593 describes the upgrade procedure. But if Sophos UTM is installed directly with version 9.5x (no upgrade from 9.x) it should be running as 64 bit module. How can I check what version of Postgres is running and if it's 32 or 64 bit?

Before I set this thread as solved I will check if the issue is coming back again after some days.

Schöne Arbeit !

Check the version with:

grep POSTGRES_ARCH /var/storage/pgsql92/postgres.default

You should get POSTGRES_ARCH="64"

Taken from the KB article Sophos UTM: Upgrading PostgreSQL to 64-bit on UTM 9.5.

Cheers - Bob

Hi all,

after one week with a new installed HA cluster of my virtual Sophos UTM environment I can confirm that everything is working fine. There are no issues and no error messages anymore and I get my daily executive reports.

And I didn't have to upgrade the postgres database because it's installed as 64 bit version.

Hi all,

after one week with a new installed HA cluster of my virtual Sophos UTM environment I can confirm that everything is working fine. There are no issues and no error messages anymore and I get my daily executive reports.

And I didn't have to upgrade the postgres database because it's installed as 64 bit version.

Hi all,

today I updated my VMware ESXi hosts to 6.7. For this I shut down the slave node within the management UI of the UTM. This worked as it should and I was able to upgrade the ESXi host. Then I waited until the slave node was back online again and then stopped the master node. From this time my internet connection didn't work correctly: Some web sites were reachable others not, i. e. HTTPS sites and it wasn't possible to download files. And there was no possibility to get access to the management UI of the UTM. After waiting a very long time I saw that the master node didn't shut down and so I stopped it with the VMware UI. Then the internet connection worked again without issues. So I did the upgrade of my other ESXi host and then started the UTM master node (preferred master). Everything seemed to work fine: The master node came back online and the faultback of the HA cluster worked.

But now, the error messages are coming back again. And the daily executive report isn't working again :-(. It Looks like there's a bug with the UTM 9.509-3 and HA cluster. Maybe there will be files corrupted when a failover of the HA cluster will happen.

I can't open a ticket because I'm a home edition user. Does someone know if this is a known bug and when it will be resolved by Sophos? Or is the only solution to install the nodes once again and hopefully never have a cluster failover or to go back to an older release?

Hi all,

I did now a factory reset of my both UTM nodes, see https://community.sophos.com/kb/en-us/115568.

The reports are working again. The question is now if I should rebuild the HA cluster or if I let the UTM run as single system until the next update? It looks like the HA mode isn't working as it should so it doesn't make sense to use it. What do you think?

Hi all,

when will an update be available for the UTM? The HA isn't working as it should! When there should be a failover the messages come back again and the failover won't happen. So this feature doesn't make sense any more.

Hello all,

yesterday I updated my Sophos UTM HA environment to version 9.510-5. And yes, HA is working again as it should! There are no error messages regarding Pop3 proxy not running, ACC device Agent not running, HA selfcheck after doing a failover and a faultback.