| Title: | VAX and Alpha VMS |
| Notice: | This is a new VMSnotes, please read note 2.1 |
| Moderator: | VAXAXP::BERNARDO |
| Created: | Wed Jan 22 1997 |
| Last Modified: | Fri Jun 06 1997 |
| Last Successful Update: | Fri Jun 06 1997 |
| Number of topics: | 703 |
| Total number of notes: | 3722 |
Along with note 82, the same customer is reviewing all of their sysgen params on all their clusters. They understand the impact and reasoning behind most of the parameters they have modified from the defaults, but several have "historical" setting which no-one recalls anything about (like why they are set at what they are for some sites). Here's a few they are asking about: 1) One site has shadow_sys_disk = 2049, while all others have it set to 1. The site with it set to 2049 seems to recall that they were told to set it to 2049 to "disable write history on hsj40s", but they don't recall why or know if it's still needed. Anyone one anything about this ? 2) Several sites have tbskipwsl = 128, while others have it at the default of 8. In reading the info in sysgen about this, it's not clear what this param really does or if setting it to 128 has any benefit for them. 3) They have SCSFLOWCUSH all over the map, ranging from the default of 1 to the max of 16. Again, each site "tweaked" this as they read DECUS articles, got advice from Col Springs, etc., but there is no consistant understanding of why you would set it higher than the default or what the overall impact is. Any advice or comments on the above is appreciated (with the one caveat that I'd like to not have "guesses")...... :^) Thanks Arlan
| T.R | Title | User | Personal Name | Date | Lines |
|---|---|---|---|---|---|
| 96.1 | BSS::JILSON | WFH in the Chemung River Valley | Tue Jan 28 1997 11:36 | 11 | |
>1) E-mailed STARS article [SHADOW] Process Hangs Accessing Shadow Set Due
To Lost I/O
>2) When a process tries to pagefault another page into its working
set, the OpenVMS Pagefault Handler searches the process' current
working set list for an EMPTY entry or a VALID entry that is not
in the Translation Buffer Cache (TB Cache). It will only search
the SYSGEN parameter 'TBSKIPWSL' working set list entries.
See [OpenVMS] Reasons for RWMPB/RWMPE States and PAGEFRAG/PAGECRIT Messages
Jilly
| |||||
| 96.2 | Start Over... | XDELTA::HOFFMAN | Steve, OpenVMS Engineering | Tue Jan 28 1997 11:36 | 7 |
Get rid of the cruft in MODPARAMS (after saving a copy somewhere safe), follow the current manuals such as the Guide to Performance Management and current system and layered product requirements, and tune to current system and application requirements. (And remember to leave some comments in MODPARAMS for the next time. :-) | |||||
| 96.3 | UTRTSC::thecat.uto.dec.com::JurVanDerBurg | Change mode to Panic! | Wed Jan 29 1997 02:15 | 26 | |
>1) One site has shadow_sys_disk = 2049, while all others have it set to 1. > The site with it set to 2049 seems to recall that they were told > to set it to 2049 to "disable write history on hsj40s", but they don't > recall why or know if it's still needed. Anyone one anything about this ? Yes, that was do disable write logging. There have been problems in the past but if everything is up to rev. you should just set it to 1. >2) Several sites have tbskipwsl = 128, while others have it at the default > of 8. In reading the info in sysgen about this, it's not clear what > this param really does or if setting it to 128 has any benefit for them. Get rid of that entry. You're unlikely to measure any performance improvements because of this, and if setting it too high it can finally hang a system. I've seen this more than once in the past. >3) They have SCSFLOWCUSH all over the map, ranging from the default of 1 > to the max of 16. Again, each site "tweaked" this as they read DECUS > articles, got advice from Col Springs, etc., but there is no consistant > understanding of why you would set it higher than the default or what > the overall impact is. Use the default for this. Jur. | |||||