The reason of print queue hanging?

[Glen Liu]

It seems that it's very common that the print queue of a HP-UX system will be hung suddently without any obvious reason and will work again after you lpshut/lpsched it. I did it last week and my client now keep asking me for a "root reason" about that. I do have no idea about it. Is it because of the volume of the printing job or a software defect? Or it's just a tricky problem based on probability, like lottery? Thanks.

[Jitendra_1]

When it seems to be hang , can you check with ps -aef| grep lpsched to see if there is any large print job going on? You can also make sure you have all the LP related patches installed on your system.

[Glen Liu]

Thanks, Hemant.

It happened last week as well as again and again. And it's gone. I get two points from your words -- large print job and lack of patches. Is my understanding correct?

[Jitendra_1]

yes those might be possible causes for your problem . Try these solutions and see if it solves the problem. You got to watch it real closely when it happens ( say through ps -ef, or in /var/spool/lp directory etc.)

[Glen Liu]

Thanks again, Hemant.

But what I need now is not solutions. I have read piles of them and thanks for those kind guys. I, or say my client, need root cause.
Thanks.

[Jitendra_1]

I understand your concerns. However to reach to the root cause , you need to gather data and evidence of the problem and try some solutions. This might not be immediate and you really need to monitor the problem closely. I am afraid i will not be able to give you the root cause here right away.

Thanks and good luck.

[Bill Hassell]

A root cause is going to be a lengthy process, simply because the print spooler can be implented is so many different ways. Printers can be connected using a parallel port, a serial port, a DTC port concentrator, HP's JetDirect card/box, some 3rd party network adapter, or sent to yet another system such as HP-UX or other Unix box or (humm) a PC-based solution like NT.

For directly connected printers (serial or parallel), the quality of the connection (ie, within supported lengths) is important. And don't overlook hardware problems if the same print queue keeps locking up.

For the DTC, there are numerous log entries to check, and the serial connection quality also applies.

For Jetdirect (as well as 3rd party print spooling solutions), the network health is critically important. A LAN can be crippled with a single misconfigured PC. HP's JetDirect software can log a lot of details about the connection quality. (this does not apply to lpr/rlp connections to JetDirect or 3rd party solutions).

When sending the print job to another system, now the other computer's patch level and loading will affect the success of the print job (one of many reasons to avoid print servers and print directly to the printer).

NT print servers come with their own problems, not the least of which is incompatibility with Unix ASCII files (the stairstep output problem).

Identifying the root cause is possible with on site consulting from HP, but it could be somewhat expensive. So I would start with patches for the LAN and the spooler, or better yet, put the GR bundle in to handle the 100+ other problems too.

[Glen Liu]

Thank you, Hament and Bill.

I will try to look more close to the problem, when it comes again, in the ways you suggested.
Thanks for your help.