Mailing List Archive

Conserver 8.x on Solaris 8...
Folks,

Before I spend too much time tracking this down, has anyone seen the following behaviour before? I'm using Conserver 8.1.5 on a Solaris 8 host connecting to several terminal servers via TCP. Every five minutes each Conserver daemon wakes up and reports the following for each console in its group... The same behaviour has been observed with older versions of Conserver 8...

[Thu May 13 15:01:29 2004] conserver (5783): ERROR: [scholes] read failure
[Thu May 13 15:01:29 2004] conserver (5783): [scholes] automatic reinitialization


A quick truss of the daemon shows that poll() is indicating 'ready to read' on a file descriptor, but that the following read() returns zero bytes. And this is flagged as an error by Conserver?

poll(0xFFBEF738, 17, 9000) = 1
fd=3 ev=POLLRDNORM rev=0
fd=6 ev=POLLRDNORM rev=0
fd=8 ev=POLLRDNORM rev=0
fd=10 ev=POLLRDNORM rev=0
fd=12 ev=POLLOUT rev=0
fd=14 ev=POLLRDNORM rev=POLLRDNORM
fd=16 ev=POLLRDNORM rev=0
fd=18 ev=POLLRDNORM rev=0
fd=20 ev=POLLRDNORM rev=0
fd=22 ev=POLLRDNORM rev=0
fd=24 ev=POLLRDNORM rev=0
fd=26 ev=POLLRDNORM rev=0
fd=28 ev=POLLRDNORM rev=0
fd=30 ev=POLLRDNORM rev=0
fd=32 ev=POLLRDNORM rev=0
fd=34 ev=POLLRDNORM rev=0
fd=36 ev=POLLRDNORM rev=0
read(14, 0xFFBEF078, 1024) = 0



Thanks,
Will.

William Charles ______________________________________________________________
Global UNIX Engineering, Asia/Pacific [/] Deutsche Bank Sydney +61 2 9258 1916



--

This e-mail may contain confidential and/or privileged information. If you are not the intended recipient (or have received this e-mail in error) please notify the sender immediately and destroy this e-mail. Any unauthorized copying, disclosure or distribution of the material in this e-mail is strictly forbidden.
Re: Conserver 8.x on Solaris 8... [ In reply to ]
On Thu, May 13, 2004 at 03:11:57PM +1000, William Charles wrote:
> Folks,
>
> Before I spend too much time tracking this down, has anyone seen the following behaviour before? I'm using Conserver 8.1.5 on a Solaris 8 host connecting to several terminal servers via TCP. Every five minutes each Conserver daemon wakes up and reports the following for each console in its group... The same behaviour has been observed with older versions of Conserver 8...
>
> [Thu May 13 15:01:29 2004] conserver (5783): ERROR: [scholes] read failure
> [Thu May 13 15:01:29 2004] conserver (5783): [scholes] automatic reinitialization
>

looks like the terminal server is timing out the console connections.
i know some have default idle-timeouts that disconnect clients after a
certain amount of inactivity. and, then again, others can be set that
way. that's my guess...a 5 minute idle-timeout.

Bryan
Re: Conserver 8.x on Solaris 8... [ In reply to ]
What type of Terminal Server?

On Thu, 2004-05-13 at 08:39, Bryan Stansell wrote:
> On Thu, May 13, 2004 at 03:11:57PM +1000, William Charles wrote:
> > Folks,
> >
> > Before I spend too much time tracking this down, has anyone seen the following behaviour before? I'm using Conserver 8.1.5 on a Solaris 8 host connecting to several terminal servers via TCP. Every five minutes each Conserver daemon wakes up and reports the following for each console in its group... The same behaviour has been observed with older versions of Conserver 8...
> >
> > [Thu May 13 15:01:29 2004] conserver (5783): ERROR: [scholes] read failure
> > [Thu May 13 15:01:29 2004] conserver (5783): [scholes] automatic reinitialization
> >
>
> looks like the terminal server is timing out the console connections.
> i know some have default idle-timeouts that disconnect clients after a
> certain amount of inactivity. and, then again, others can be set that
> way. that's my guess...a 5 minute idle-timeout.
>
> Bryan
> _______________________________________________
> users mailing list
> users@conserver.com
> https://www.conserver.com/mailman/listinfo/users
Re: Conserver 8.x on Solaris 8... [ In reply to ]
> A quick truss of the daemon shows that poll() is indicating 'ready
> to read' on a file descriptor, but that the following read() returns
> zero bytes. And this is flagged as an error by Conserver?

"zero bytes read" is how unix reports EOF.

run snoop/tcpdump/ethereal on the interface and I bet you'll see the
TS closing the connections..

- Bill
Re: Conserver 8.x on Solaris 8... [ In reply to ]
Many thanks to all those who responded. All were right in saying that the likely culprit was our Cisco terminal servers timing-out the connections after five minutes... I'm not yet sure whether our Network colleagues have changed anything, but I'm almost certain that Conserver 7.x was somewhat less sensitive to this? I certainly don't recall the logs being full of reconnection messages? Could be wrong...

Thanks Again,
Will.

William Charles ______________________________________________________________
Global UNIX Engineering, Asia/Pacific [/] Deutsche Bank Sydney +61 2 9258 1916




bryan@conserver.c
om To: users@conserver.com
Sent by: cc:
users-bounces@con Subject: Re: Conserver 8.x on Solaris 8...
server.com


13/05/2004 22:39






On Thu, May 13, 2004 at 03:11:57PM +1000, William Charles wrote:
> Folks,
>
> Before I spend too much time tracking this down, has anyone seen the following behaviour before? I'm using Conserver 8.1.5 on a Solaris 8 host connecting to several terminal servers via TCP. Every five minutes each Conserver daemon wakes up and reports the following for each console in its group... The same behaviour has been observed with older versions of Conserver 8...
>
> [Thu May 13 15:01:29 2004] conserver (5783): ERROR: [scholes] read failure
> [Thu May 13 15:01:29 2004] conserver (5783): [scholes] automatic reinitialization
>

looks like the terminal server is timing out the console connections.
i know some have default idle-timeouts that disconnect clients after a
certain amount of inactivity. and, then again, others can be set that
way. that's my guess...a 5 minute idle-timeout.

Bryan
_______________________________________________
users mailing list
users@conserver.com
https://www.conserver.com/mailman/listinfo/users





--

This e-mail may contain confidential and/or privileged information. If you are not the intended recipient (or have received this e-mail in error) please notify the sender immediately and destroy this e-mail. Any unauthorized copying, disclosure or distribution of the material in this e-mail is strictly forbidden.
Re: Conserver 8.x on Solaris 8... [ In reply to ]
I've had issues where a firewall/router will kill a connection if no
data goes through after a length of time.

On Thu, 2004-05-13 at 20:42, William Charles wrote:
> Many thanks to all those who responded. All were right in saying that the likely culprit was our Cisco terminal servers timing-out the connections after five minutes... I'm not yet sure whether our Network colleagues have changed anything, but I'm almost certain that Conserver 7.x was somewhat less sensitive to this? I certainly don't recall the logs being full of reconnection messages? Could be wrong...
>
> Thanks Again,
> Will.
>
> William Charles ______________________________________________________________
> Global UNIX Engineering, Asia/Pacific [/] Deutsche Bank Sydney +61 2 9258 1916
>
>
>
>
> bryan@conserver.c
> om To: users@conserver.com
> Sent by: cc:
> users-bounces@con Subject: Re: Conserver 8.x on Solaris 8...
> server.com
>
>
> 13/05/2004 22:39
>
>
>
>
>
>
> On Thu, May 13, 2004 at 03:11:57PM +1000, William Charles wrote:
> > Folks,
> >
> > Before I spend too much time tracking this down, has anyone seen the following behaviour before? I'm using Conserver 8.1.5 on a Solaris 8 host connecting to several terminal servers via TCP. Every five minutes each Conserver daemon wakes up and reports the following for each console in its group... The same behaviour has been observed with older versions of Conserver 8...
> >
> > [Thu May 13 15:01:29 2004] conserver (5783): ERROR: [scholes] read failure
> > [Thu May 13 15:01:29 2004] conserver (5783): [scholes] automatic reinitialization
> >
>
> looks like the terminal server is timing out the console connections.
> i know some have default idle-timeouts that disconnect clients after a
> certain amount of inactivity. and, then again, others can be set that
> way. that's my guess...a 5 minute idle-timeout.
>
> Bryan
> _______________________________________________
> users mailing list
> users@conserver.com
> https://www.conserver.com/mailman/listinfo/users
>
>
>
>
>
> --
>
> This e-mail may contain confidential and/or privileged information. If you are not the intended recipient (or have received this e-mail in error) please notify the sender immediately and destroy this e-mail. Any unauthorized copying, disclosure or distribution of the material in this e-mail is strictly forbidden.
>
>
> _______________________________________________
> users mailing list
> users@conserver.com
> https://www.conserver.com/mailman/listinfo/users
Re: Conserver 8.x on Solaris 8... [ In reply to ]
I think one temp solution may be to change the keepalive on the
conserver server to something less than 5 minutes. That may keep it
going.

On Thu, 2004-05-13 at 20:42, William Charles wrote:
> Many thanks to all those who responded. All were right in saying that the likely culprit was our Cisco terminal servers timing-out the connections after five minutes... I'm not yet sure whether our Network colleagues have changed anything, but I'm almost certain that Conserver 7.x was somewhat less sensitive to this? I certainly don't recall the logs being full of reconnection messages? Could be wrong...
>
> Thanks Again,
> Will.
>
> William Charles ______________________________________________________________
> Global UNIX Engineering, Asia/Pacific [/] Deutsche Bank Sydney +61 2 9258 1916
>
>
>
>
> bryan@conserver.c
> om To: users@conserver.com
> Sent by: cc:
> users-bounces@con Subject: Re: Conserver 8.x on Solaris 8...
> server.com
>
>
> 13/05/2004 22:39
>
>
>
>
>
>
> On Thu, May 13, 2004 at 03:11:57PM +1000, William Charles wrote:
> > Folks,
> >
> > Before I spend too much time tracking this down, has anyone seen the following behaviour before? I'm using Conserver 8.1.5 on a Solaris 8 host connecting to several terminal servers via TCP. Every five minutes each Conserver daemon wakes up and reports the following for each console in its group... The same behaviour has been observed with older versions of Conserver 8...
> >
> > [Thu May 13 15:01:29 2004] conserver (5783): ERROR: [scholes] read failure
> > [Thu May 13 15:01:29 2004] conserver (5783): [scholes] automatic reinitialization
> >
>
> looks like the terminal server is timing out the console connections.
> i know some have default idle-timeouts that disconnect clients after a
> certain amount of inactivity. and, then again, others can be set that
> way. that's my guess...a 5 minute idle-timeout.
>
> Bryan
> _______________________________________________
> users mailing list
> users@conserver.com
> https://www.conserver.com/mailman/listinfo/users
>
>
>
>
>
> --
>
> This e-mail may contain confidential and/or privileged information. If you are not the intended recipient (or have received this e-mail in error) please notify the sender immediately and destroy this e-mail. Any unauthorized copying, disclosure or distribution of the material in this e-mail is strictly forbidden.
>
>
> _______________________________________________
> users mailing list
> users@conserver.com
> https://www.conserver.com/mailman/listinfo/users
Re: Conserver 8.x on Solaris 8... [ In reply to ]
That's an idea. I can also confirm that the observed behaviour is indeed due to the terminal server having (recently) been configured to drop idle connections...

Thanks,
Will.

William Charles ______________________________________________________________
Global UNIX Engineering, Asia/Pacific [/] Deutsche Bank Sydney +61 2 9258 1916




cfowler@outpostse
ntinel.com To: William Charles/Sydney/DBAustralia/DeuBa@DBAustralia
Sent by: cc: users-bounces@conserver.com, users@conserver.com,
users-bounces@con bryan@conserver.com
server.com Subject: Re: Conserver 8.x on Solaris 8...


14/05/2004 11:10
Please respond to
cfowler






I think one temp solution may be to change the keepalive on the
conserver server to something less than 5 minutes. That may keep it
going.

On Thu, 2004-05-13 at 20:42, William Charles wrote:
> Many thanks to all those who responded. All were right in saying that the likely culprit was our Cisco terminal servers timing-out the connections after five minutes... I'm not yet sure whether our Network colleagues have changed anything, but I'm almost certain that Conserver 7.x was somewhat less sensitive to this? I certainly don't recall the logs being full of reconnection messages? Could be wrong...
>
> Thanks Again,
> Will.
>
> William Charles ______________________________________________________________
> Global UNIX Engineering, Asia/Pacific [/] Deutsche Bank Sydney +61 2 9258 1916
>
>
>
>
> bryan@conserver.c
> om To: users@conserver.com
> Sent by: cc:
> users-bounces@con Subject: Re: Conserver 8.x on Solaris 8...
> server.com
>
>
> 13/05/2004 22:39
>
>
>
>
>
>
> On Thu, May 13, 2004 at 03:11:57PM +1000, William Charles wrote:
> > Folks,
> >
> > Before I spend too much time tracking this down, has anyone seen the following behaviour before? I'm using Conserver 8.1.5 on a Solaris 8 host connecting to several terminal servers via TCP. Every five minutes each Conserver daemon wakes up and reports the following for each console in its group... The same behaviour has been observed with older versions of Conserver 8...
> >
> > [Thu May 13 15:01:29 2004] conserver (5783): ERROR: [scholes] read failure
> > [Thu May 13 15:01:29 2004] conserver (5783): [scholes] automatic reinitialization
> >
>
> looks like the terminal server is timing out the console connections.
> i know some have default idle-timeouts that disconnect clients after a
> certain amount of inactivity. and, then again, others can be set that
> way. that's my guess...a 5 minute idle-timeout.
>
> Bryan
> _______________________________________________
> users mailing list
> users@conserver.com
> https://www.conserver.com/mailman/listinfo/users
>
>
>
>
>
> --
>
> This e-mail may contain confidential and/or privileged information. If you are not the intended recipient (or have received this e-mail in error) please notify the sender immediately and destroy this e-mail. Any unauthorized copying, disclosure or distribution of the material in this e-mail is strictly forbidden.
>
>
> _______________________________________________
> users mailing list
> users@conserver.com
> https://www.conserver.com/mailman/listinfo/users

_______________________________________________
users mailing list
users@conserver.com
https://www.conserver.com/mailman/listinfo/users





--

This e-mail may contain confidential and/or privileged information. If you are not the intended recipient (or have received this e-mail in error) please notify the sender immediately and destroy this e-mail. Any unauthorized copying, disclosure or distribution of the material in this e-mail is strictly forbidden.