ASE 15.0.2 && periodic message: kernel Cannot read, host process disconnected

[rebelde]

Hello,

We have an installation on Linux SLES10 with ASE 15.0.2 and a lot of ASE
clients connected. All day long with a periodic of exactly 5 minutes we
see in the ASE log file the following messages(no other messages):

...
00:00000:00063:2010/12/29 00:00:21.23 kernel Cannot read, host process disconnected: spid: 63
00:00000:00029:2010/12/29 00:05:21.75 kernel Cannot read, host process disconnected: spid: 29
00:00000:00065:2010/12/29 00:10:22.70 kernel Cannot read, host process disconnected: spid: 65
00:00000:00087:2010/12/29 00:15:32.29 kernel Cannot read, host process disconnected: spid: 87
00:00000:00091:2010/12/29 00:20:32.29 kernel Cannot read, host process disconnected: spid: 91
00:00000:00020:2010/12/29 00:25:32.40 kernel Cannot read, host process disconnected: spid: 20
00:00000:00088:2010/12/29 00:30:32.31 kernel Cannot read, host process disconnected: spid: 88
00:00000:00108:2010/12/29 00:35:32.38 kernel Cannot read, host process disconnected: spid: 108
00:00000:00093:2010/12/29 00:40:32.39 kernel Cannot read, host process disconnected: spid: 93
00:00000:00055:2010/12/29 00:45:32.31 kernel Cannot read, host process disconnected: spid: 55
00:00000:00058:2010/12/29 00:50:32.51 kernel Cannot read, host process disconnected: spid: 58
00:00000:00045:2010/12/29 00:55:32.42 kernel Cannot read, host process disconnected: spid: 45
00:00000:00084:2010/12/29 01:00:32.70 kernel Cannot read, host process disconnected: spid: 84
00:00000:00060:2010/12/29 01:05:32.80 kernel Cannot read, host process disconnected: spid: 60
00:00000:00070:2010/12/29 01:10:32.82 kernel Cannot read, host process disconnected: spid: 70
...

and as well in our ASE client messages we have no hints about a mal
function;

What could cause the above messages? Thanks

matthias

[Rob V]

Post by rebelde
Hello,
We have an installation on Linux SLES10 with ASE 15.0.2 and a lot of ASE
clients connected. All day long with a periodic of exactly 5 minutes we
...
00:00000:00063:2010/12/29 00:00:21.23 kernel Cannot read, host process disconnected: spid: 63
00:00000:00029:2010/12/29 00:05:21.75 kernel Cannot read, host process disconnected: spid: 29
00:00000:00065:2010/12/29 00:10:22.70 kernel Cannot read, host process disconnected: spid: 65
00:00000:00087:2010/12/29 00:15:32.29 kernel Cannot read, host process disconnected: spid: 87
00:00000:00091:2010/12/29 00:20:32.29 kernel Cannot read, host process disconnected: spid: 91
00:00000:00020:2010/12/29 00:25:32.40 kernel Cannot read, host process disconnected: spid: 20
00:00000:00088:2010/12/29 00:30:32.31 kernel Cannot read, host process disconnected: spid: 88
00:00000:00108:2010/12/29 00:35:32.38 kernel Cannot read, host process disconnected: spid: 108
00:00000:00093:2010/12/29 00:40:32.39 kernel Cannot read, host process disconnected: spid: 93
00:00000:00055:2010/12/29 00:45:32.31 kernel Cannot read, host process disconnected: spid: 55
00:00000:00058:2010/12/29 00:50:32.51 kernel Cannot read, host process disconnected: spid: 58
00:00000:00045:2010/12/29 00:55:32.42 kernel Cannot read, host process disconnected: spid: 45
00:00000:00084:2010/12/29 01:00:32.70 kernel Cannot read, host process disconnected: spid: 84
00:00000:00060:2010/12/29 01:05:32.80 kernel Cannot read, host process disconnected: spid: 60
00:00000:00070:2010/12/29 01:10:32.82 kernel Cannot read, host process disconnected: spid: 70
...
and as well in our ASE client messages we have no hints about a mal
function;
What could cause the above messages? Thanks
matthias

It just means the client for that spid has disappeared without politely
disconnecting and informing the ASE server. So the ASE server at some
point finds out that the client is no longer there, and then reports
this message.
As far as ASE is concerned, there's nothing wrong or harmful about this.
It can happen because of many reasons, like pulling the power plug on
the PC where the client runs. But given that you say it happens at
regular intervals, maybe there's a KEEPALIVE setting set to 5 minutes
somewhere?

It may be interesting to figure out if those disconnected spids have
something in common (like: are they all coming from the same host?).
When you set this config param:

sp_configure 'log audit logon success', 1

... then ASE will write the username, spid and IP address of the client
host to the ASE errorlog for every new connection. That way you can
perhaps gain some more insights.

HTH,

Rob V.
-----------------------------------------------------------------
Rob Verschoor

Certified Sybase Professional DBA for ASE 15.0/12.5/12.0/11.5/11.0
and Replication Server 15.0.1/12.5 // TeamSybase

Author of Sybase books (order online at www.sypron.nl/shop):
"Tips, Tricks& Recipes for Sybase ASE" (ASE 15 edition)
"The Complete Sybase ASE Quick Reference Guide"
"The Complete Sybase Replication Server Quick Reference Guide"

***@NO.SPAM.sypron.nl | www.sypron.nl | Twitter: @rob_verschoor
Sypron B.V., The Netherlands | Chamber of Commerce 27138666
-----------------------------------------------------------------

[rebelde]

...

Post by Rob V

Post by rebelde
00:00000:00060:2010/12/29 01:05:32.80 kernel Cannot read, host process
disconnected: spid: 60
00:00000:00070:2010/12/29 01:10:32.82 kernel Cannot read, host process
disconnected: spid: 70 ...
and as well in our ASE client messages we have no hints about a mal
function;
What could cause the above messages? Thanks
matthias

It just means the client for that spid has disappeared without politely
disconnecting and informing the ASE server. So the ASE server at some
point finds out that the client is no longer there, and then reports
this message.
As far as ASE is concerned, there's nothing wrong or harmful about this.
It can happen because of many reasons, like pulling the power plug on
the PC where the client runs. But given that you say it happens at
regular intervals, maybe there's a KEEPALIVE setting set to 5 minutes
somewhere?
It may be interesting to figure out if those disconnected spids have
something in common (like: are they all coming from the same host?).
sp_configure 'log audit logon success', 1
... then ASE will write the username, spid and IP address of the client
host to the ASE errorlog for every new connection. That way you can
perhaps gain some more insights.

The ASE-clients are UNIX application servers running on the same host as
ASE. The clients of these servers are running on PC connecting through TCP
ports to the application servers. In their log files are similar
connect/disconnect requests all 5 minutes and all from one specific PC.
Will check what's doing this PC, perhaps some kind of regular interval port
scan or port monitor.

Thanks for the hint regarding 'log audit logon...'

matthias

[rebelde]

Post by rebelde
The ASE-clients are UNIX application servers running on the same host as
ASE. The clients of these servers are running on PC connecting through TCP
ports to the application servers. In their log files are similar
connect/disconnect requests all 5 minutes and all from one specific PC.
Will check what's doing this PC, perhaps some kind of regular interval
port scan or port monitor.

To close this thread: A Nagios port scanner/monitor was causing this.

matthias

[s***@gmail.com]

Post by rebelde

Post by rebelde
The ASE-clients are UNIX application servers running on the same host as
ASE. The clients of these servers are running on PC connecting through TCP
ports to the application servers. In their log files are similar
connect/disconnect requests all 5 minutes and all from one specific PC.
Will check what's doing this PC, perhaps some kind of regular interval
port scan or port monitor.

To close this thread: A Nagios port scanner/monitor was causing this.
matthias

Noticed your quote about a network monitor causing Sybase client disconnects. We are new to Linux and was running a load test with 200 users and hit a wall on a new Linux server were we started loosing client connections to Sybase with the same error message.

As a follow-up question, I was wondering if you had to stop port monitoring or was it a setting in Linux; I am not sure if we have any port monitoring running, but if we did, how can you tell and shut it down?

The client disconnect issue was not an issue on our old HP-UX servers... We ran the same stress test on those servers and then attempted to run the same on the new Linux servers, but the test failed as we started loosing clinets connections, with the same error message and the only clue to an issue.

-Thanks.