CSWS 2.1 hangs

[Dieter Rossbach]

I use CSWS 2.1 on alpha VMS 8.3, all patches intalled.
After a couple of hours, I get more and more processes in "COM" state and only a shutdown and restart helps.
The error log shows a couple of messages like
[Thu Jun 14 19:59:06 2007] [error] (6)no such device or address: error assigning
channel to control mailbox APACHE$SWS_CONTROL_MBX_5\n

Google was no help.

Any Ideas?

Dieter

[Karl Rohwedder]

Just a wild guess, have you checked some quotas e.g. FILLM,BYTLM,CHANNELCNT et al. ?

regards kalle

[Bill Hall]

Dieter,

VMS 8.3 allows mailbox unit numbers above 32767. We had one application "break" because of this.

Check the value of the APACHE$*_CONTROL_MBX logical name periodically. You may be able to verify whether or not CSWS can handle mailbox unit numbers above 32767.

We logged a call last winter on this and got a patch for the mailbox driver from VMS Engineering. I don't think the patch has been released in an official ECO as of yet.

Bill

[Bill Hall]

Dieter,

I thought I should check the versions of CSWS we are running as we may run into the same problem in September when we will complete our VMS 8.3 upgrades. We have not upgraded any of our clusters that run CSWS yet.

We are running CSWS 2.1-1 on VMS 8.2. I checked the CSWS website and found that there is an update to 2.1-1 that fixes two problems, one of which is described as:

Zero-length files cause server processes to become CPU-bound and hang.

Could this be your problem? Are you running the 2.1 or 2.1-1 with or without Update 01?

Bill

[Dieter Rossbach]

Bill,

my apache version is 2.1-1

I don't think it is a quota problem:

Process Quotas:

CPU limit: Infinite Direct I/O limit: 200
Buffered I/O byte count quota: 248064 Buffered I/O limit: 500
Timer queue entry quota: 999 Open file quota: 271
Paging file quota: 1482912 Subprocess quota: 20
Default page fault cluster: 64 AST quota: 996
Enqueue quota: 9991 Shared file limit: 0
Max detached processes: 0 Max active jobs: 0

Accounting information:
Buffered I/O count: 2925 Peak working set size: 20864
Direct I/O count: 18879791 Peak virtual size: 218912
Page faults: 1462 Mounted volumes: 0
Images activated: 4
Elapsed CPU time: 0 01:31:18.30


and

"APACHE$SWS_CONTROL_MBX" = "MBA21867:

is not to high

Dieter

[Bill Hall · Accepted solution]

Dieter,

Do you have CSWS 2.1-1 update 1 that is referenced on this page installed?

http://h71000.www7.hp.com/openvms/products/ips/apache/csws_patches.html#sws

Bill

[Dieter Rossbach]

Ok, looks good so far.

Thank your for your help. Update was installed but not active. Apache$shutdown does not (at least on my system) remove the installed images. An $ Image remove on all apache stuff, a $deassign/system/exec on some apache logicals an a apache$restart and here we go: good performance, stable apache processes.

Thanks again

Dieter

[Bill Hall]

Dieter,

Thanks for the update on this problem. I can check off CSWS our VMS 8.3 upgrade checklist.

Bill