Cache Buffers at 85%

[bayeon]

Hi

I have a Poweredge 4200 running on Netware 3.12, I believe someone put in quite a few patch. Applications we are running is written in Foxpro, written in dos base.
The guy that maintain the server left us high and dry.
Will need any type of help I can get from you guys/gals, to keep the server alive, while we slowly migrate over to a new system.

First question would be... My Cache Buffers is running at 85% all the time 453 out of 512MB, even after I reboot. I am a windows guy, so I don't really know where to start.

2nd question, what kind of maintenance can I do to keep the server running at a decent condition? Defragment? clean up the temp files (how do I do that?)? Is there such thing as chkdsk?

3rd, I have an steady stream of error message, "..Insufficient directory space (or memory) to preserve sys:005f001.q during erase..."

Thanks,

 

[Provogeek]

You can check the patch level of your server by downloading the CONFIG.NLM from the support site. This will create a CONFIG.TXT in the SYSTEM directory on the SYS volume. Then you can also download the Config Reader that will tell you whats up with your server by reading the CONFIG.TXT file. If you can find someone that can sell it to you you can install the 3.2 enhancment pack that is just all the patches for 3.12 in package.

NOTE: 3.x is no longer supported, so Novell will not help you. If you need help, look for a local partner to help you with your NetWare server until you migrate off of it.



1. Were are you looking for this cache buffer? 85% normally a good thing.

2. NetWare is much different that Windows, tehre is no defrag or chkdisk utility. You don't need to worry about defragging the hard drive, there just isn't any tool for 3.x that will do that (it's old and out dated). Just run a PURGE on the network drive from a workstation about once a month to clean out deleted files. Other than that, the file system basicly cleans it's self up.

3. Not sure on this

Provogeek

Saving the world, one network at a time

 

[JockMullin]

Hi, Bayeon

Item 1. Provogeek is right, if you have 85% memory allocated to cache that means only 15% is tied up by the OS. Novell dynamically allocates all available memory to cache. How much memory does the server have?

Item 2. Well, there is the CHKVOL utility in 3.12. It will tell you how much space is allocated/available on a volume. From a workstation map a drive letter to the volume (say W and then enter command CHKVOL W:. This will give you a summary of disk space used and available and how much is occupied by deleted files and how much is available to you based on your logon id/rights. It is a reporting utility only - it does not change anything.

Unless purge deleted files is set to on Novell automatically keeps all deleted files until the space is required for new allocations in which case they get purged on an oldest first basis. This is a great feature if you have the memory to support it because you can recover deleted files using the salvage utility.

This may be related to item 3. It looks like the server is saying it cannot keep print queue entries either because it is out of memory or directory entries. Deleted files are still in the directory until purged. There is a max limit on the number of directory entries per volume. Also Novell caches all directories, so if you don't have enough memory or if the volume has a lot of files then you may get this error.

Most people don't put application files on the SYS volume, so it is not all that volatile except for queue entries, which could explain why you are getting those errors.

Provogeek's suggestion about doing a purge all is good - that will remove all deleted files and save directory entries and consequently cache space. If you don't need to be able to salvage files you can set it so deleted files are purged immediately upon deletion.

Suggestions:

Do a CHKVOL on the SYS volume and see where you stand re disk space.

Check to see if you are getting any server low on memory messages. If not then likely the problem is directory size.

Do a disk cleanup on the SYS volume and delete obsolete/unnecessary files or move them to another volume or put them into a .zip archive to save directory entries.

Do a purge all on the sys volume and see if that gets rid of the insufficient directory entries messages. This will only help if the CHkVOL shows space being occupied by deleted files.

Jock

 

[bayeon]

Thanks for all the info guys!

 

[iolair]

All good advice from the other responders. It is also possible to check a disk in Netware using VREPAIR, but it requires you unmount the volume before you run that check. I only do that when the server gets turned off unexpectedly, like when there is a power failure over the weekend, and I wasn't here to perform a down and exit. VRepair will repair those disk errors. But, it's usually NOT necessary to do anything to a Netware Volume.

As for the cache buffers - I'd *love* to have 85% - I usually have around 78 - 79%, and I never have any problems at all. I understand that Novell does have documentation about what is considered "normal" cache buffers. Adding RAM will raise your free buffers.

The third question sounds like a print queue that has been corrupted. If you go to PCONSOLE, you can match up the directory name 005f001.q to the actual print queue. You might be able to delete the queue and recreate it, but, again, you'll have to be able to stop all print jobs going to that queue while you work on it. Then again, it may have nothing to do with printing at all. Hope that helps.

Iolair MacWalter