NLM1.5 Index problems (Networking Products)
At 26 MAY 1998 05:13:50AM Marshall Gray wrote:
We just installed version 1.5 of the NLM on our server last week are beginning to experience index errors. We cannot rebuild them with the NLM running. We are running Novell 4.11; AREV 3.12. From the documentation we have been changing the parameters without success. The index is on a warehouse file of 14,400 records with 14 columns being indexed. The problem arises when indexing the long description column (long text field) in the table. Our current settings are as follows:
LHIPXTSR.EXE /C#120 /R:100 /P
LH /S:2 (OTHERS ARE DEFAULT)
LHIPXSER /P:512 (OTHERS ARE DEFAULT)
When trying to rebuild the index we get the fatal error FS466 "no server response". After offloading the NLM, rebuilding the index, and reloading the NLM again, we still get requests from the application to rebuild the index again. The file in question also has two MFS's on it.
Any suggestions would be appreciated.
At 27 MAY 1998 10:44PM Marshall Gray - progress update wrote:
We continued to play with the parameters and changed the LH /S:# value from 2 to 64 and the LHIPXSER /P:# value from 512 to 4000. We are now able to rebuild indexes with the NLM running. However, we still get repeated requests from the application to rebuild this index. Has anyone else experienced index related problems after installing the NLM?
Thanks to John for clarifying the ctrl-alt-del question. It is as you said, the files close on the server in about 10 minutes.
At 29 MAY 1998 12:55AM Mark Martin wrote:
Hi Marshall,
You wrote:
We continued to play with the parameters and changed the LH /S:# value from 2 to 64 and the LHIPXSER /P:# value from 512 to 4000. We are now able to rebuild indexes with the NLM running. However, we still get repeated requests from the application to rebuild this index. Has anyone else experienced index related problems after installing the NLM? Thanks to John for clarifying the ctrl-alt-del question. It is as you said, the files close on the server in about 10 minutes.
It sounds like 1 of 2 things:
1) Someone is not using the NLM driver at their workstation or a workstation is not recognizing the NLM and it is doing indexing.
Under this scenario, you should set the ServerOnly=1 in the REVPARAM file for each directory the Linear-Hash (REV12345.LK and .OV files are stored at). This way you will find the culprit.
2) Someone is shutting down their system (ctrl-alt-del or powering off or losing a connection to the server) while running the application and it is processing indexes.
Under this scenario, you should set up a dedicated indexer and turn off indexing on all workstations – both background indexing as well as update indexes before query in the environments for your users. To accomplish this, set up a low end workstation, log it into the application with a special environment that has the station continuously updating indexes and that is all it does.
I hope this helps.
Mark Martin
At 31 MAY 1998 02:11AM Marshall Gray wrote:
Thank you, Mark, for the advice. After reading all the other postings here regarding index problems we opted to try the dedicated indexing machine also and so far it seems to be working.
Many thanks.
At 12 JUN 1998 04:34AM Sergey Yazikov wrote:
We installed the indexer, however, we still get repeated requests from the application to rebuild this index, it is just not too often. Also I think the suggestion to use a low end machine was not very good, because on my ManageWise network dashboard I see that machine is making the highest utilization among others, and is on the top of the Bytes Out/sec list.
Any clue?