Backup Central

Hi everybody,

is there a recommended value for the client inactivity timeout?
I think I´ve read once that a bad value is one of several reasons for RPC-Errors.

Reason why I´m asking is: Some nights we get a big lot of RPC errors, especially after our three big groups start. Each of these groups contains 50 to 60 clients (windows and linux), they are scheduled with a time difference of 30 minutes.
Inactivity timeout is set to 60.

The message looks like that:
nsrindexd RPC error: Aborting client connection from "IP-address"; why = A connection attempt failed because the connected party did not properly respond after a period of time, or established connection failed because connected host has failed to respond.

Server is W2K3, Networker Version of Server and Storage nodes ist 7.6.3.3

Regards, Christian

Hi Christian,

I have seen similar instances of this where the fault lies with the
backup server DB unable to handle all the communication.
If you monitor the load on the backup server resources does it use lots
of memory and many disk I/O's??
I spread the load of the backups by creating more groups and putting a
maximum of 20 to 25 clients per group.
if you can do this also stagger the start times to prevent the memory
usage and disk I/O from peaking.
I hope this helps.
Andy


On 05/06/2012 13:47, Christian Jurecka wrote:
Hi everybody,

is there a recommended value for the client inactivity timeout?
I think I´ve read once that a bad value is one of several reasons for RPC-Errors.

Reason why I´m asking is: Some nights we get a big lot of RPC errors, especially after our three big groups start. Each of these groups contains 50 to 60 clients (windows and linux), they are scheduled with a time difference of 30 minutes.
Inactivity timeout is set to 60.

The message looks like that:
nsrindexd RPC error: Aborting client connection from "IP-address"; why = A connection attempt failed because the connected party did not properly respond after a period of time, or established connection failed because connected host has failed to respond.

Server is W2K3, Networker Version of Server and Storage nodes ist 7.6.3.3

Regards, Christian

Hi,
Have you monitored the backup server resorces including network utilization and ports ? perhaps your server is busy, staggering start time and smaller groups or adding a storage node may be the answer, follow the best practices for the networker.
Another temporary fix may be to add addtional bandwith on the server ( network & or Memory\CPU).
Hope that helps
Fazil Saiyed


Date: Tue, 5 Jun 2012 14:51:09 +0100
From: farske10 < at > HOTMAIL.COM
Subject: Re: [Networker] inactivity timeout question
To: NETWORKER < at > LISTSERV.TEMPLE.EDU

Hi Christian,

I have seen similar instances of this where the fault lies with the
backup server DB unable to handle all the communication.
If you monitor the load on the backup server resources does it use lots
of memory and many disk I/O's??
I spread the load of the backups by creating more groups and putting a
maximum of 20 to 25 clients per group.
if you can do this also stagger the start times to prevent the memory
usage and disk I/O from peaking.
I hope this helps.
Andy


On 05/06/2012 13:47, Christian Jurecka wrote:
Hi everybody,

is there a recommended value for the client inactivity timeout?
I think I´ve read once that a bad value is one of several reasons for RPC-Errors.

Reason why I´m asking is: Some nights we get a big lot of RPC errors, especially after our three big groups start. Each of these groups contains 50 to 60 clients (windows and linux), they are scheduled with a time difference of 30 minutes.
Inactivity timeout is set to 60.

The message looks like that:
nsrindexd RPC error: Aborting client connection from "IP-address"; why = A connection attempt failed because the connected party did not properly respond after a period of time, or established connection failed because connected host has failed to respond.

Server is W2K3, Networker Version of Server and Storage nodes ist 7.6.3.3

Regards, Christian


=

Hi,
I know that there is a way to increase the the value for the TCP/IP timeout in miliseconds. This is a registry setting. I cannot remember the name, can possibly
find by Google or at Microsoft.
On the other side it is a good hint to decrease the number of clients in the groups, properly the workload on the nic is to high.
Regards Rainer

Am 05.06.2012 um 16:30 schrieb Bombay4u None:

Hi,
Have you monitored the backup server resorces including network utilization and ports ? perhaps your server is busy, staggering start time and smaller groups or adding a storage node may be the answer, follow the best practices for the networker.
Another temporary fix may be to add addtional bandwith on the server ( network & or Memory\CPU).
Hope that helps
Fazil Saiyed


Date: Tue, 5 Jun 2012 14:51:09 +0100
From: farske10 < at > HOTMAIL.COM
Subject: Re: [Networker] inactivity timeout question
To: NETWORKER < at > LISTSERV.TEMPLE.EDU

Hi Christian,

I have seen similar instances of this where the fault lies with the
backup server DB unable to handle all the communication.
If you monitor the load on the backup server resources does it use lots
of memory and many disk I/O's??
I spread the load of the backups by creating more groups and putting a
maximum of 20 to 25 clients per group.
if you can do this also stagger the start times to prevent the memory
usage and disk I/O from peaking.
I hope this helps.
Andy


On 05/06/2012 13:47, Christian Jurecka wrote:
Hi everybody,

is there a recommended value for the client inactivity timeout?
I think I´ve read once that a bad value is one of several reasons for RPC-Errors.

Reason why I´m asking is: Some nights we get a big lot of RPC errors, especially after our three big groups start. Each of these groups contains 50 to 60 clients (windows and linux), they are scheduled with a time difference of 30 minutes.
Inactivity timeout is set to 60.

The message looks like that:
nsrindexd RPC error: Aborting client connection from "IP-address"; why = A connection attempt failed because the connected party did not properly respond after a period of time, or established connection failed because connected host has failed to respond.

Server is W2K3, Networker Version of Server and Storage nodes ist 7.6.3.3

Regards, Christian

I would be of the openion that increasing timeout in milisecounds on the server side may have triveal affects , if RPC services are bing affected , it may not respond for longer then few secounds or longer, so fixing the root cause may be your preferred solution.
Hope that helps
Fazil Saiyed


Subject: Re: [Networker] inactivity timeout question
From: rethmeier.rainer < at > web.de
Date: Tue, 5 Jun 2012 20:32:03 +0200
To: NETWORKER < at > LISTSERV.TEMPLE.EDU; bombay4u < at > HOTMAIL.COM

Hi,
I know that there is a way to increase the the value for the TCP/IP timeout in miliseconds. This is a registry setting. I cannot remember the name, can possibly
find by Google or at Microsoft.
On the other side it is a good hint to decrease the number of clients in the groups, properly the workload on the nic is to high.
Regards Rainer

Am 05.06.2012 um 16:30 schrieb Bombay4u None:

Hi,
Have you monitored the backup server resorces including network utilization and ports ? perhaps your server is busy, staggering start time and smaller groups or adding a storage node may be the answer, follow the best practices for the networker.
Another temporary fix may be to add addtional bandwith on the server ( network & or Memory\CPU).
Hope that helps
Fazil Saiyed


Date: Tue, 5 Jun 2012 14:51:09 +0100
From: farske10 < at > HOTMAIL.COM
Subject: Re: [Networker] inactivity timeout question
To: NETWORKER < at > LISTSERV.TEMPLE.EDU

Hi Christian,

I have seen similar instances of this where the fault lies with the
backup server DB unable to handle all the communication.
If you monitor the load on the backup server resources does it use lots
of memory and many disk I/O's??
I spread the load of the backups by creating more groups and putting a
maximum of 20 to 25 clients per group.
if you can do this also stagger the start times to prevent the memory
usage and disk I/O from peaking.
I hope this helps.
Andy


On 05/06/2012 13:47, Christian Jurecka wrote:
Hi everybody,

is there a recommended value for the client inactivity timeout?
I think I´ve read once that a bad value is one of several reasons for RPC-Errors.

Reason why I´m asking is: Some nights we get a big lot of RPC errors, especially after our three big groups start. Each of these groups contains 50 to 60 clients (windows and linux), they are scheduled with a time difference of 30 minutes.
Inactivity timeout is set to 60.

The message looks like that:
nsrindexd RPC error: Aborting client connection from "IP-address"; why = A connection attempt failed because the connected party did not properly respond after a period of time, or established connection failed because connected host has failed to respond.

Server is W2K3, Networker Version of Server and Storage nodes ist 7.6.3.3

Regards, Christian




=

I think you need to find the Windows equivalent of the follwing
Solaris commands:

/usr/sbin/ndd -set /dev/tcp tcp_conn_req_max_q 1024
/usr/sbin/ndd -set /dev/tcp tcp_conn_req_max_q0 4096

This is probably some registry key you need to tweak. Maybe you can find
it in Networker's Performance Tuning guide.



On 06/05/2012 03:47 PM, Christian Jurecka wrote:
Hi everybody,

is there a recommended value for the client inactivity timeout?
I think I´ve read once that a bad value is one of several reasons for RPC-Errors.

Reason why I´m asking is: Some nights we get a big lot of RPC errors, especially after our three big groups start. Each of these groups contains 50 to 60 clients (windows and linux), they are scheduled with a time difference of 30 minutes.
Inactivity timeout is set to 60.

The message looks like that:
nsrindexd RPC error: Aborting client connection from "IP-address"; why = A connection attempt failed because the connected party did not properly respond after a period of time, or established connection failed because connected host has failed to respond.

Server is W2K3, Networker Version of Server and Storage nodes ist 7.6.3.3

Regards, Christian


--

-- Yaron.