exchange 2007,

Are the basic resources ok on the server? Specially the free hard drive space of the volumes where Exchange componentes are located?

yes cpu and ram are ok

error is 17020 error

can you please elaborate as every once in a week back pressure kicks in

Hi,

here is a good article on back pressure, for exchange 2007. As schnellsolutions said, you need to keep an eye for CPU, RAM and HDD. You will find values in this article.

http://www.msexchange.org/articles-tutorials/exchange-server-2007/management-administration/understanding-back-pressure-feature-exchange-server-2007.html 

Regards,
Ivan.

we are good at cpu, ram, disk, my question which nobody is clearly answering is:

if there is 17020 error showing up and causing transport service to shut down, CAN IT CREATE BACK PRESSURE?

when back pressure kicks in it acuses lsass.exe, w3wp.exe to really spike?

How many exchange servers you have?

Remember that you can have a domino effect on many aspects. For example, the transport service shuts down... it means that when your users try to send emails, they will stay on the mailboxes submission queues pending of being delivered to the hub transport queue. It means that the value of this counter will increase and eventually the high value on it is going to affect in many ways. Another example, if there is more than one exchange server, Exchange1 will try to send a message to Exchange2, but Exchange 2 is with the transport service down, it means that Exchange1 will increment its local queue waiting for a established service on Exchange1... and obviously it will increment the message queue on Exchange2 and can also activate back pressure.

There are different scenarios where it can easily happen, there fore we can really assume that such fails can create back pressure problems.

i have following:

1) CAS and hub server
2) 2 mailbox server
3) 1 exchange server which HAS EMC only no exchange services running.

So ... Definetely a transport service error can affect your queues and consequently may activate backPressure.

It would be nice to complete the previously listed reviews about performance counters, antivirus exclusions, service timeout value, and queue database corruption.

increase default timeout value for the ServicesPipeTimeout value in the registry to avoid the problem:

how this will help ?

If it applies that one of your services require a long time to start, then you can specify to the Service Control Manager to extend the time for the service start before it categorizes it as timeout. The example on the article presents that if your "Windows Trace Session Manager service" does not start in 30 seconds by default, then the system will timeout it as far as that threshold was reached. If for some reason your server is requiring more time than the default for starting a service (maybe because of performance issues that you cannot repair), then you can increment that threshold.

The example where it is incremented from 30seconds to 1 minute the threshold for timeout: https://support.microsoft.com/en-us/kb/839803

One last questions

thanks for all your suggestion , I will be closing this question after my below questin

is there any way I can repair queue folder without running eseutil command

can increasing version buckets of high , medium, normal to 500 help? right now they are at 200

can I perform offline defragmentation of mail.que database

yes you can, exactly in the same way than it is done with a normal .edb file. The only difference is the purpose of the file and the extension.

Be aware that you will need to stop the transport service, and that transport server is not going to be routing emails while the process is done.

On another hand... if you are at a point where your queues are empty on that server (You can check it with Get-Queue from the EMS), then you can stop the service, move your transport service files somewhere and start back the service again, which will just create a new queue database file starting from zero. Maybe it will impact less your service, however, double check that your queue does not have any pending message before going with this option.

so just move the queue database which is queue folder, but how inany way that will help if there is corruption

If there is corruption inside your Queue database, and you move it. Then when the service start a new one will be created (without corruption), as far as it is a brand new file.

how long it will take to do everything, half hr max.?

yeah, we can tell that half hour maximum.

In reality, if your queues are clean... the time that it takes you to stop the service, move or rename the files and start the service back is just a very few minutes.

By the way, experience this process in a lab first, so you will cover it completely before making it on production.

Another detail is the idea of moving or renaming the original files in order that you stay with these files as a backup.