Hi,
I have a strange behavor with Exchange 2013, and I need some extra help to identify the culprit.
For the technical context, we have a pair of Exchange 2013 servers, and a Netscaler V12 appliance which acts a load balancer.
Each of is server (Let's say SRV01 and SRV02) is hosting a mailbox database as master (Let's say DB01 and DB02), and the other one as passive.
Since a few weeks, we noticed that application pool "MSExchangeSyncAppPool" fails on the server hosting DB02 (SRV02).
As a matter of fact, service is degraded, and some of our mobile devices are not syncing anymore.
A further analysis showed that, on the server hosting DB02 active database (SRV02), IIS Active connections were growing over the time, but were never ended. The only way to solve the problem was to recycle the application pool, and IIS connections were released.
We used this article as a basis for investigation : https://support.microsoft.com/en-us/help/4456227/activesync-clients-cannot-connect-or-synchronizing-delayed-in-exchange
We noticed that :
- Indeed ActiveSync Application pool memory usage went from 200Mo to 5Go, on SRV02, in a few days.
- W3SVC_W3WP/Active Requests counter for the MSExchangeSyncAppPool is growing, on SRV02, until recycling it.
We first act on "\Exchange Server\V15\FrontEnd\HttpProxy\sync\web.config", in order to increase value "HttpProxy.ConcurrencyGuards.TargetBackendLimit". We upgraded it from 150 to 5000.
At first, it seemed to bring a better situation. But the problem is now happening between day 2 and 4 after the recycle.
We dig a bit further with powershell scripts, in order to properly investigate ActiveSync connections from our mobiles.
Here is the one we used : https://gallery.technet.microsoft.com/scriptcenter/ActiveSyncReport-script-a2417a84#content
With this, we generate a report for a 3 days timelapse.
We noticed that the first user in the list has 18000 hits recorded for the timelapse, alongside 16000 pings.
If we examine this user's activity during the day, we see :
- During 23 of the 24 hours daytime, we have a stable amount of hits/pings (25 to 75)
- For the remaining hour, we reach a peak (From 2249 hits / 2220 pings to 7402 hits / 7377 pings)
These users are connecting to Exchange with Microsoft Outlook app from Google PlayStore. Nothing unofficial.
So you understand us when we deduced that this was quite unusual to see high numbers in short period of time.
According to our research, it may seem that ActiveSync ping requestes are stranded.
We can always recycle the application pool manually or with a scheduled task. But we'd rather like to understand why endpoints are sudenly acting like this.
Any help ?
Thanks a lot.
I have a strange behavor with Exchange 2013, and I need some extra help to identify the culprit.
For the technical context, we have a pair of Exchange 2013 servers, and a Netscaler V12 appliance which acts a load balancer.
Each of is server (Let's say SRV01 and SRV02) is hosting a mailbox database as master (Let's say DB01 and DB02), and the other one as passive.
Since a few weeks, we noticed that application pool "MSExchangeSyncAppPool" fails on the server hosting DB02 (SRV02).
As a matter of fact, service is degraded, and some of our mobile devices are not syncing anymore.
A further analysis showed that, on the server hosting DB02 active database (SRV02), IIS Active connections were growing over the time, but were never ended. The only way to solve the problem was to recycle the application pool, and IIS connections were released.
We used this article as a basis for investigation : https://support.microsoft.com/en-us/help/4456227/activesync-clients-cannot-connect-or-synchronizing-delayed-in-exchange
We noticed that :
- Indeed ActiveSync Application pool memory usage went from 200Mo to 5Go, on SRV02, in a few days.
- W3SVC_W3WP/Active Requests counter for the MSExchangeSyncAppPool is growing, on SRV02, until recycling it.
We first act on "\Exchange Server\V15\FrontEnd\HttpProxy\sync\web.config", in order to increase value "HttpProxy.ConcurrencyGuards.TargetBackendLimit". We upgraded it from 150 to 5000.
At first, it seemed to bring a better situation. But the problem is now happening between day 2 and 4 after the recycle.
We dig a bit further with powershell scripts, in order to properly investigate ActiveSync connections from our mobiles.
Here is the one we used : https://gallery.technet.microsoft.com/scriptcenter/ActiveSyncReport-script-a2417a84#content
With this, we generate a report for a 3 days timelapse.
We noticed that the first user in the list has 18000 hits recorded for the timelapse, alongside 16000 pings.
If we examine this user's activity during the day, we see :
- During 23 of the 24 hours daytime, we have a stable amount of hits/pings (25 to 75)
- For the remaining hour, we reach a peak (From 2249 hits / 2220 pings to 7402 hits / 7377 pings)
These users are connecting to Exchange with Microsoft Outlook app from Google PlayStore. Nothing unofficial.
So you understand us when we deduced that this was quite unusual to see high numbers in short period of time.
According to our research, it may seem that ActiveSync ping requestes are stranded.
We can always recycle the application pool manually or with a scheduled task. But we'd rather like to understand why endpoints are sudenly acting like this.
Any help ?
Thanks a lot.