Full_Name: Guillermo Payet Version: 1.2 OS: Red Hat Linux 5.1 URL: Submission from: (NULL) (208.242.218.40) Hello, We've been running OpenLDAP 1.0.3 over dn 1.85 for a while, with no major probelms. Last Friday, I downloaded 1.2 and DB 2.3.16. After some initial trouble compiling the new version (problems with config.cache), I got a system up and running. Now slapd will run fine for a few hours, and then it will deadlock. I haven't been able to trigger the problem at will. Sometimes it hangs, sometimes it doesn't, and it seems to be independent of load. Once it hangs I have to killall -9 slapd. I thought it might be a threading issue, but if I pound the server hard with a few clients at the same time, it will run fine, and then it will freeze a few hours later with only one user. How should I approach this? what loglevel should I use to obtain a meaningful log file? thanks --Guillermo Payet
Froze again!! This time I managed to get some more info... We were running slapd with errorlevel 4. Now every time I attempt a call to the server, it logs: ldap_pvt_thread_create failed (1) If I attempt a "killall slapd", slapd logs: slapd shutting down - waiting for 10 threads to terminate and just waits... only way to kill it is with "kill -9" I have gdb attached to slapd, but in my ignorance, I have no clue as to what to do next! any hints? --G At 09:17 AM 2/18/99 , you wrote: >Full_Name: Guillermo Payet >Version: 1.2 >OS: Red Hat Linux 5.1 >URL: >Submission from: (NULL) (208.242.218.40) > > > >Hello, > >We've been running OpenLDAP 1.0.3 over dn 1.85 for a while, >with no major probelms. Last Friday, I downloaded 1.2 and >DB 2.3.16. After some initial trouble compiling the new >version (problems with config.cache), I got a system up and >running. > >Now slapd will run fine for a few hours, and then it will >deadlock. I haven't been able to trigger the problem at will. >Sometimes it hangs, sometimes it doesn't, and it seems to >be independent of load. Once it hangs I have to killall -9 >slapd. > >I thought it might be a threading issue, but if I pound the >server hard with a few clients at the same time, it will run >fine, and then it will freeze a few hours later with only one >user. > >How should I approach this? what loglevel should I use to >obtain a meaningful log file? > > thanks > > --Guillermo Payet > > > ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Guillermo Payet voice: 408.466.0700 O C E A N G R O U P http://www.oceangroup.com ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
moved from Incoming to Software Bugs
The problem was due to the buggy phtreads implementation in glibc inb Red Hat linux 5.1. (http://www.openldap.com/faq/index.cgi?file=92&showLastModified=1) I upgraded glibc, recompiled, and that fixed it... --G ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Guillermo Payet voice: 831.466.0700 O C E A N G R O U P http://www.oceangroup.com ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
changed notes changed state Open to Closed
gpayet@oceangroup.com wrote: > > The problem was due to the buggy phtreads implementation in glibc inb Red Hat > linux 5.1. (http://www.openldap.com/faq/index.cgi?file=92&showLastModified=1) > > I upgraded glibc, recompiled, and that fixed it... > > --G > > ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ > Guillermo Payet voice: 831.466.0700 > O C E A N G R O U P http://www.oceangroup.com > ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ -- peter@netline.net.uk Peter Duffy Unix Developer Netline (UK) Ltd.
We're using openldap 1.2.0 and Redhat 5.2 (with glibc-2.0.7-29, including libpthread-0.7.so, apparently at the same level as described.) We're still seeing an intermittent deadlock problem which seems to match this description. (Sorry about the previous blank followup to this: I hit the send button prematurely...) gpayet@oceangroup.com wrote: > > The problem was due to the buggy phtreads implementation in glibc inb Red Hat > linux 5.1. (http://www.openldap.com/faq/index.cgi?file=92&showLastModified=1) > > I upgraded glibc, recompiled, and that fixed it... > > --G > > ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ > Guillermo Payet voice: 831.466.0700 > O C E A N G R O U P http://www.oceangroup.com > ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Regards, Peter -- peter@netline.net.uk Peter Duffy Unix Developer Netline (UK) Ltd.
Vendor bug. see follup #2.