The client read timeout and client connect timeout can go on masters or medias. They go on anything that will talk to a client.

WHY ON EARTH do you keep on repeating the SAME questions?

Please see this post (in case you forgot):

https://www.veritas.com/connect/forums/client-read-and-client-connect-time-out

Please also study the process flow diagram in Appendix A of the Troubleshooting Guide.

If you understand comms and data flow between media server and client, you will understand why timeouts need to go on media servers.

Adding to mariannes outstanding post ...

Why do you have timeouts on clients - they do nothing why are they being added.

Are you adding timeouts just because you can ?  This is not the way to administrate a backup environment.

Untill you are very very experienced, I do not recommend changing any settings like timeouts unless you have a problem.

However, perhaps you did have some issue and you added the timeouts to the master or media, and , the clients without realising they do not go there.

When you add setting chanages, you must add only ONE at a time, and see what the effect is, and, if it does not do what you want remove it.

For example, perhaps you have some issue where you have to add a client connect timeout to the media server to fix it.

If you :

1.  Add the client connect timeout to the media server, and test, you will see that it resolved the issue - job done.

2.  If you add the timeout to the client and test you will see it doesn't fix the issue, so, you remove it and try the timeout on the media server - which would be the correct solution.

3.  If you add he timeout to the media server and the client at the same time and then test, you will see tha th issue is solved BUT, only because you have added the setting to the media server.  The setting on the client is doing nothing and sholdn't be there, but as you made two changes at once, this is hidden.

Do you see why you should only make one change at once, and, why only if you have an issue ???

You can also get in the situation where you make 2 (or more) changes where each of the changes would  break the system but with the same symptoms.  Even when troubleshooting, we only like to make 1 change at a time, so if you have mutiple changes that have been made that break the system with the same symptoms they are very hard to find

I saw a case some time ago - system was in a right mess - wouldn't run at all, but 'no changes had been made'.

It took about a week to get the system running .  In fact so many changes had been made that caused the same or similar issues, that it was almost impossibe to find them.  

It turned out that many of these changes had been made for no real reason, just people 'playing' - eventually it broke the system, and then more changes were made that made the situation even more worse.

Martin

Have a look at a backup job in Activity Monitor.

You will notice a line saying 'Connecting'.
This is when the Media server initiates a connection request.
Media server expects a 'Connect Back' from the client before 'CLIENT_CONNECT_TIMEOUT' expires.
When Client connects back to media server, you will see 'Connected'. This normally happens within a few seconds.

So, on your media server where Connect Timeout is 20001 seconds (5.55 HOURS) this is the time that your media server will wait for the 'Connected'. If there is some sort of problem on the client (we don't know what) the media server will sit and 'wait' for more than 5 hours before connection request will time out.

Same for read timeout - media server will wait for client to send data. If client does not send any data or stops before backup is finished (for example, client has crashed or network card died), media server will wait for data to come through until Read Timeout arrives.
In your case, again more than 5 hours, wasting media server resources....
And then people log calls with Symantec because backup jobs hang......

The defaut Connect and Read timeout is 300 seconds (5 minutes). This is enough in MOST cases.
As Martin explained - ONLY increase timeouts when specific problems are experienced.

It is 'normal' to increase these timeouts when clients with large databases are being backed up using the database agents (e.g. Oracle). The databases utilities sometimes take a long time to start the backup.
Here we normally increase the timeouts on the media server to 1800 (30 minutes).

WHAT DOES IT MEAN?

It means that someone in your environment has been PLAYING nicely without understanding what he or she was doing!
 

OK, so that is useful information - where you actually having some problem ?

Normally, when you adjust a timeout, you just increase it a little bit ... not sudenly jump to several hours.

It is possible to completely bring down a system by makeing changes that aren't very carefully planned.

As I mentioned, if multiple changes are made - it can take days or even weeks to fix, during which time backups are not running ...

M

Thanks martin.

I want to know whats the meaning if we add these on media servers and master server?

On client i understand.

I have below entries on my media servers..

CLIENT_NAME = usa0300ux868

CLIENT_CONNECT_TIMEOUT = 20001

CLIENT_READ_TIMEOUT = 20001

SERVER_CONNECT_TIMEOUT = 20001

BPSTART_TIMEOUT = 30002

BPEND_TIMEOUT = 30001

Below entries on my master server:

CLIENT_CONNECT_TIMEOUT = 900

BPSTART_TIMEOUT = 900

 

What does this mean?

Good point Marianne.

so what exactly is difference between ctimeout settings on both client and media server.

ok so timeout settings need to be on the media server and not on te client.

Thanks ALL.