We are suddenly getting an ever-increasing number of open files each day until the server crashes at around 10,000 files.
Articles and Tips: qna
01 Jan 1998
Q.
Dear Ab-end: On a NetWare 4.10 server (a Compaq Proliant computer with a disk array), we are suddenly getting an ever-increasing number of open files each day until the server crashes at around 10,000 files (usually once per week). The first 6,000 occur with the first backup, then the number increments about 700-1000 daily. I have been scouring the web as well as Novell documentation and I can't find anything even remotely similar to this problem. We use Arcserve 5.01g (about to upgrade to 6.1), but the only change we made was the evening before the open files issue began to occur. I had installed a number of Compaq's NLMs in SYS:SYSTEM. One of the modules failed to see the changer, so I was forced to return the system back to the way it was. Do you have any suggestion where I might even start to research this problem further?
-- Files Everywhere and Not a One that Closes
A.
Dear Files: It sounds like there could be a problem with the version of the backup software you are running. The best way to determine if the problem lies with NetWare or with Arcserve is to wait until you are running Arcserve again. When it has a few thousand open files, unload Arcserve and see if it releases the files. You will probably see a number of messages run across the console screen telling you that ARCSERVE.NLM was unloaded without releasing all of its resources.
You may even have the server "abend" with a message something to the effect of "NLM UNLOADED WITHOUT RELEASING ALL OF ITS SEMAPHORES." If this occurs, you should run a core dump and send it to Novell so we can verify what is happening. (See the AppNote entitled "Troubleshooting Server Problems Using the ABEND.LOG File and Memory Images (Core Dumps)" in the October 1997 issue of AppNotes.)
It sounds like the backup software may be having a problem with opening files during the backup procedure (which you verify through the Arcserve log files). When the backup process finishes, it does not handle the open files error properly and therefore leaves these files open. After you perform this test, we will have further information to work on.
If the problem lies with NetWare, there's a good chance the CLIB that Arcserve has been using was overcopied with a different version of CLIB that is not properly handling open files, or it has simply been corrupted. Try reloading the CLIB library software that came with NetWare 4.10 or with ARCSERVE 5.01g (if there is one).
(Editor's note: the sender of this question has since upgraded his version of Arcserve and this problem went away. But the steps given are still valid in narrowing down similar problems.)
* Originally published in Novell AppNotes
Disclaimer
The origin of this information may be internal or external to Novell. While Novell makes all reasonable efforts to verify this information, Novell does not make explicit or implied claims to its validity.