Level of Parallelism exceeded

Question

Running Legato 7.1 server on Solaris 8.
Running Netware client 4.22 on Netware 6.5, SP3

Have two tape libraries, each with two drives.

We dedicate one library (via tape pools used in it) to backing up VMS, UNIX & Microsoft. The other library is dedicated to Netware clients.

The Netware groups run at 8pm, 12am, 2AM & 4AM, and each have 2-3 clients.

Periodically one of the clients in one of the Groups will fail, with the message "current request exceeds parallelism". I've seen this occur, even when other Netware jobs are done.

I've bumped up the "parallelism" field in the client's setup to 6 or higher, but it didn't fix the issue.

Any ideas?

dpinink_silva · Answer

The devices also have a parallelism field, and also the server. What's this value for you devices?

ble1 · Answer

Daniel,Devices don't have parallelism field in 7.1.x (or 7.2.x). They have target session and that is different from parallelism in NetWorker world (which exists for client, group and server in those versions). Max number of sessions (which is what parallelism is in NetWorker world) for devices exists since 7.3.(jukebox parallelism I didn't mention as it is not related to save stream mechanism and its allocation).Message was edited by: hcrvelin

ble1 · Answer

Charles,Do yourself a favour and if you are really running 7.1 go to 7.1.4 as first step. That has nothing to do with your query (even in 7.1.3 savegrp parallelism issue was resolved, but that doesn't seem to be your issue from what you have said so far), but this is very much recommended. List of patches:- For UNIX http://web1.legato.com/infodev/publications/NetWorker/UNIX/7.1.4/uxfbl_714.pdf - For Windows http://web1.legato.com/infodev/publications/NetWorker/WINDOWS/7.1.4/winfbl_714.pdf(bare in mind that this is not complete list, but rather list of addresses escalations only)The message 'Current request exceeds parallelism' I saw only once when the client tried to generate a number of save streams which was not supported by XBSA interface of the database application (Exchange namly). The parallelism should decreased then and not increased.Assumptions: If I understand correctly what you have said, this is reported by client and not server (given that savegrp reports it I would assume this is passed by save process). Thus I would assume your server parallelism is big enough to receive all your streams (otherwise you would see messages in daemon.log lile 're-setting overall number of authorized save sessions, changing value from (number)...). Nevertheless, I would follow that path and do some client check to see if system logs do not indicate anything wrong and then I would decrease client parallelism and increase server parallelism.

DavidHampson · Answer

If your server parallelism is set to 12 and client parallelism is set to 6 then you may only be able to back up 2 clients at a time... It really depends on the volumes of data on each of the clients, the amount of disks on them and how long we expect their backups to take.

cgoff1 · Answer

Thank you for your posting. I am running 7.1 (build 230). Currently the Server parallelism is at 12 (too high or ok?).

The parallelism for the Netware client was at 6.

The error was indicated by looking at "details" on the Savegrp icon. Not sure if this is reported by client or server.

Also, since we are on maintenance, should we go to the latest and greatest 7.21? Or as you advised, 7.1.4?

Thanks again.

ble1 · Answer

Since you are in 7.2 follow that path - go to 7.2.1.

As for server parallelism - it really depens, but no matter what version you have you can go up to (at least) 32. Let us assume you have fine server, target sessions are as they are and just increase it for certain value (to 16 for example). Also, for NetWare client decrease it to 4 and see if that brings any changes.

NetWorker

Level of Parallelism exceeded

Was this post helpful?