Hi. We had the opposite problem with the same sympthons: the same bad performance but free memory.
Same as you, I would like to know how to rule the amount of memory each uv-process takes. We did "ps -el | grep uv" in order to see the memory usage. (The "SZ" column gives you the number of 4 kb pages that every uv-process uses). Finally we discovered that uv-processes didn't ask for more memory, so that we increased the HP-tunable: "dbc_max_pct" to 70%, in order to use the free memory as disk caché. But the main benefit was given by resizing some files with heavy overflow. HTH. -- Augusto 2010/6/8 u2ug <[email protected]> > We started running into a situation recently where a box running > universe 10.2.11 on hpux 11.31 started to have severe performance > problems, zero response, processes that normally take seconds to run > taking hours etc. > > In Top we could see that the reported free memory was averaging about > 28M at times dropping to less than 5M and that the vhand process was > chewing up a most of the cpu. With the free memory that low, vhand > hogging the system makes sense. Why the memory is that low is the > question. > > The machine has 4G of memory, right after reboot before anyone but me > gets on the system top shows us at 3G free. If we just let the system > sit, free memory will sit happily at 3G. As soon as we start using > universe, the free memory starts to fall. Logging out all universe > processes and even bringing universe down has no effect on free memory - > I would have expected the memory to be freed up once the universe > processes and in particular universe itself was terminated but that's > not what we are seeing. > > We were in the process of setting up a new box physically the same as > the old box but running uv 10.3.6 on hpux 11.3.1. So I am looking at > this problem now on that box in isolation. We were hoping that this > would turn out to be a hardware issue but we are seeing the same problem > on this new box as well. > > This is out of my realm of experience and I am stumped as to where to go > next. I have the system guy checking with hp , my guess is they are > going to point the finger at uv. I am going to check with our var - I'm > not overly optimistic there either. > > Any suggestions ? > - Hpux/uv tunables we should be looking at > - can we tell universe to restrict the amount of physical memory > that it uses ? > - Or maybe this is a garbage collection issue ? > - universe or hpux being overly aggressive about caching > - any hpux utilities that we can use to see exactly where the memory is > being used ? > - .... ? > > > Gerry > > > > > -----Original Message----- > From: [email protected] > [mailto:[email protected]] On Behalf Of Martin Scholl > Sent: June 8, 2010 09:41 AM > To: 'U2 Users List' > Subject: [U2] [Unidata-Windows] Can login, but not execute any command > > Upfront: The problem was solved by re-installing UniData > > My client had UniData 7.2.3 on Windows within his domain. He rebooted > all > servers within his company over the weekend. Yesterday UniData suddenly > didn't work anymore. I could log in to the database, but any command, > whether Write, Select or catalogued subroutine would stop dead in its > track. > There was no error message, no indication what could have gone wrong. > UniObjects had no time-out, the application hung for hours. > > I telneted into the database and many users got the unknown user error, > I > just found one user that allowed me to log in. I ran a few commands and > that > worked. > > My hunch is that a Windows update changed NT-Authentication and UniData > didn't like it. Poking through the logs, I see in the Windows security > log > two entries for every UniData login, one success and one failure: > The first a success audit: > A trusted logon process has registered with the Local Security > Authority. > This logon process will be trusted to submit logon requests. > Logon Process Name: \udapi_server.exe > > Then a failure audit: > The logon to account: HHC-Unidata > by: MICROSOFT_AUTHENTICATION_PACKAGE_V1_0 > from workstation: HORIZONSERV4 > failed. The error code was: 3221225572 > > The error code means: 32212255720 The specified user does not exist. > > This behavior started already last year, and even now that everything > works > again. > > Does anyone have ideas or similar experiences? > > > Martin Scholl > 18910 New Hampshire Ave > Brinklow, MD 20862 > Phone: 301-924-5537 > Cell: 301-613-9572 > [email protected] > www.HIPAAsuite.com > > > _______________________________________________ > U2-Users mailing list > [email protected] > http://listserver.u2ug.org/mailman/listinfo/u2-users > > > _______________________________________________ > U2-Users mailing list > [email protected] > http://listserver.u2ug.org/mailman/listinfo/u2-users > _______________________________________________ U2-Users mailing list [email protected] http://listserver.u2ug.org/mailman/listinfo/u2-users
