Backup process went to RWAST mode

[Santhosh_5]

Hi,

any idea how to stop RWAST MODE PROCESSES. One of my backup process went to RWAST MODE. I tried to stop the id but it has failed. Also cross checked the BYTLM parameters and found that normal.

2040507E ABS_6FA5C64_2 RWAST 4 210715 0 00:00:31.17 944 485 S
20404CC9 SCHED#_238 HIB 6 28284 0 00:00:14.90 1192 1001

One of the HP web site suggested to restart the system for deleting this process. Is it correct? My server is production servers and we cannot restart the server just like that. any idea pls.

Thanks in advance.

[Antoniov.]

Hi Santhosh,
did you read this
http://h71000.www7.hp.com/wizard/wiz_7120.html

Antonio Vigliotti

[Volker Halle]

Santhosh,

once a process is hung in RWAST (waiting for some resource, which typically becomes available again via delivery and processing of an AST), it may be nearly impossible to manually delete this process, if the operation causing the process to go into RWAST does not somehow complete.

First steps for troubleshooting this type of problem:

- go into SDA on the running system with
$ ANAL/SYS

- set process context to the hanging process
SDA> SET PROC/INDEX=2040507E

- search for BUSY channels with
SDA> SHOW PROC/CHANNEL

- look at the call stack of the hung process
SDA> CLUE CALL

- look at the JIB (for quotas/limits)
SDA> SHOW PROC
SDA> READ SYSDEF
SDA> FORMAT JIB

These steps may give you an idea, what this process is really waiting for. It may - or may not - be possible to resolve the RWAST hang by trying to make the resource available, but you first need to find out, what the process is waiting for...

Volker.

[Bojan Nemec]

Hi,

There is a page relating RWASTs:

http://www.yrl.co.uk/phil/vms/rwast.html

Bojan

[Ian Miller.]

You may be able to look at this processing using AMDS/Availability Manager. This will tell you if its a quota problem and parhaps allow you to fix it.

If it is a quota problem then it shows that the relative values of the username used for the backup job are incorrect.

With a backup job an other posibility is that is an device i/o problem. If the backup is to a tape drive then sometimes turing the tape drive off and on will reset the drive and complete the I/O that backup is waiting for.

Is the disk that is being backed up ok?

[Lawrence Czlapinski]

Santhosh,
1. As Jan suggested if AVAIL_MAN security has a version on the tartget system which has C enabled (if the version doesn't mention C (for control) in the comments, you will need a later version of AVAIL_MAN), you can use AVAIL_MAN to look at and fix (adjust) process quotas.
2. What are the quotas for BYTLM and TQLEM?
How much of these quotas is being used?
http://h71000.www7.hp.com/wizard/wiz_7120.html
http://forums1.itrc.hp.com/service/forums/questionanswer.do?threadId=69931
Lawrence

[Jan van den Ende]

Santhosh,

even before trying to kill the RWAST process, first answer a more important question:
Is this process using any resource in a painfull way?

If this process is NOT keeping any files or devices out of service, you MIGHT also consider to just forget about it until the next reboot.


Obviously, keeping the tapedrive allocated counts as painfull in this context, and then, for now, you are limited to a reboot.
Yes, I wrote "for now". Just today at the Technical Update Days at was announced that starting V8.2 (december) there is a command to free the device!!!.
I do not yet have the presentations available, and I cannot remember just now whether this also has a TIMA kit for previous versions coming.

Cheers.

Have one on me.

Jan

[Andy Bustamante]

How to free the tape drive if
you leave the process alone until the next reboot.

For a SCSI tape drive, identify a free SCSI id. Reset the tape drive to the new scsi id. Use "mc sysman io autoconfig/log" to identify and configure the tape drive.

You'll need to modify your backup job to use the new tape drive. You should still use Availabiltiy Manager to check quotas, you don't want to keep repeating this.

[Uwe Zessin]

Jan,
you cannot 'kill' a process in RWAST state. If my memory is correct, the process is waiting at IPL #2 - the same IPL at which the SKAST to indicate a process termination request would have to be delivered.

[Jan van den Ende]

Uwe,

maybe I should not have written "trying to kill", but if you read carefully again, you will notice that my advice is quite the other way: IF it is not blocking anything you need, just forget about it. The alternative indeed IS a reboot! Then again, there ARE some RWAST reasons that can be tricked to proceed just far enough to reach the request to commit suicide (which is just another way of describing STOP/ID=..)

Cheers

Have one on me

Jan