MID server unable to subscribe to AMB channel

[mark69]

We use a MID server, but the MID server log keeps giving these warnings:

ECCQueueMonitor.40 WARNING *** WARNING *** Reconnecting AMB channel..

ECCQueueMonitor.40 WARNING *** WARNING *** Unable to subscribe to AMB channel: /mid/server/77a7ba104fed1200b92af1728110c7dd

From the documentation I can understand the function of the AMB channel, but I cannot find any info on what could be causing connection to the channel to fail.

We use another MID server connected to a customers instance, and this one does not produce these warnings.

Both are hosted from the same machine, so the problem seems to be instance related.

Also, the MID server producing these warnings has a memory leak, and will eventually die with an out of memory message even though it is not actively used.

The other MID server has no memory leakage problems.

Both are fresh installs.

Any ideas?

The instances are different versions by the way. Helsinki for the one producing the warnings, Fuji for the other one.

[mark69]

My dev instance was upgraded to Istanbul Patch 1, but unfortunately it did not solve our problems.

So I am still getting the AMB channel warnings, still seeing the threads count increase and the memory usage also increases but not as quickly as it used to.

Regards,

Mark

[ralphschoch]

I was having a similar issue, even with Istanbul at patch 3, but only in one DMZ. After forcing an upgrade on the MID server even though it showed that it did not need it, the problem went away. Just 1 more thing you (or anyone else with the same problem) might try.

[mark69]

I am now on Istanbul patch 5 hotfix 1, and I removed and reinstalled the MID server. Unfortunately the problem still remains. Thread count and memory keeps increasing. But good to hear it solved the problem in your case

[Nithya V Sivara]

HI Mark and others,

We recently upgraded our Test instance to Istanbul and we faces the same issue 'Unable to subscribe to AMB Channel' and we raised a HI ticket to Servicenow. They mentioned that the issue could be because of the Http Client Thread count issue and suggested us to go for Istanbul patch 6 but after applying the patch we still see the same issue.

Is your issue got fixed ? This completely stops us from running the discovery. Any suggestion from you all will help us solve the issue.

[mark69]

Hi Nithya,

You are saying you cannot use the mid server at all?

In my case the mid server does still work, maybe not as quick/respnsive as it used to.

As a workaround to the thread count (and memory usage) increasing over time I have created a script that restarts the mid servers. I call this script from the Windows task scheduler so this is done automatically. Until ServiceNow comes up with a solution this works for us.