Answered Hung Appserver..

rzr

Member
Hello,

The DBAs use the OEM interface to check status of the app servers. This does show error state or offline. Unfortunately, this still doesn’t show a state where the app server is online but hung.

Is there any other way to identify hung AppServer state?
 

rzr

Member
What are the identifiable symptoms of a "hung" AppServer?
There's isn't one common symptom. The one's I noticed is no updates to appsever log files. The DBA team restarts the appserver and I see activity in the appserver log files. Also the ABL client get's an error "Unable to connect to xxxx appserver..."
 

rzr

Member

This is the top section of my screen...If there were a hung appserver where would that show here?

Auto Interval Rate JSON 20771 5712 0.832 ProTop Version 3.3nx 2016/12/20 16:36:38
ptxxxx 0 0 /db/xxxx xxxx

Hit% 100.00 Commits: 0 DB UpTime: 12:49 Oldest TRX: 00:00:00 Connections: 17
Log Reads: 215 Undos: 0 Bkp Age: 18:50 Curr BIClstr: 0 -n % 1
OS Reads: 0 Lock Tbl HWM: 113 Oldest BIClstr: 0 Brokers: 2
Rec Reads: 46 Lock Tbl% 0.00 New RM: 0 Num BIClstrs: 0 4gl Servers: 0
LogRd/RecRd: 4.70 Curr # Locks: 0 From RM: 0 BI MB Used: 0 SQL Servers: 2
Log Writes: 0 Modified Bufs: 11 From Free: 0 Curr AI Extent: Disabled 4gl Clients: 8
OS Writes: 0 Examined: 0 Curr AI Seq#: 0 SQL Clients: 0
Rec Creates: 0 IO Response: 13.00 Front2Bk: 0 Empty AI Exts: 0 App Server: 1
Rec Updates: 0 BogoMIPS: 3.66 Remove Lk: 0 Full AI Exts: 0 Web Speed: 0
Rec Deletes: 0 Locked AI Exts: 0 BIW/AIW/WDOG: 1 0 1
Rec Locks: 3 Notes: 0 0 APW Writes: 0 AI Mgmt: 0
Rec Waits: 0 BIW/AIW Write% 0 0 APW Write% 0 APWs: 3
Idx Blk Spl: 0 Writes to Log: 0 0 Bufs Scanned: 23 Local: 2
Resrc Waits: 0 BIW/AIW Writes: 0 0 APW Scan Wrts: 0 Remote: 0
Latch Waits: 0 Partial Buf Wr: 0 0 APW Q Wrts: 0 Batch: 6
pica Used: 0 Busy Buf Waits: 0 0 Chkpt Q Wrts: 0 TRX: 1
pica Used% 0.00 Empty Buf Wts: 0 0 Flushed Bufs: 0 Blocked: 0
 

TomBascom

Curmudgeon
Nope. That is the default main dashboard. It does show you, over on the right side, how many app server sessions are running (unless you're on an ancient, obsolete and unsupported release of Progress).

For app servers you need to first define them (because everyone has different name for their app servers). To do that you edit etc/appsrv.cfg. There are instructions inside the .cfg file but, to summarize, if you want to monitor an app server named "xyzzy" add the following line:

apsv xyzzy "asbman -i &1 -q"

You do NOT need to define lines for the admin server, name server or webspeed unless you have those components installed and want to monitor them (I mostly wouldn't bother with admin server and name server).

Then restart ProTop and type "@". You can dismiss the UIO, Table & Index activity displays if you have a small window. (Type the "u", "t" and "i" commands to do that.)

FWIW -- I doubt that your app servers are "hung". It's more likely that all of your agents are busy executing requests and that new connections are timing out waiting for an available agent. But that should be obvious once you start monitoring.
 
  • Like
Reactions: rzr

cj_brandt

Active Member
The broker log should show an error in the log if there are no available app server agents. It is a good idea to scan that log for errors. Do you allow the broker to spawn more agents if needed ?

asbman -query will show the status of the app server agents. You need to see what state the app server agents are in. Are they all in SENDING ? If so look at their activity in the database to see what they are doing.
 

LarryD

Active Member
If the appserver is hanging due to all of the app server agents stuck in SENDING, there is a OE fix patch for 10.2B08 (it's either 58 or 60 I believe) and one for 11.3.3 (if I remember it's something like 38 but not really sure). For other versions I have no idea. I believe it was permanently fixed in one of the later 11.6.x versions.
 
Top