hi, i've a debian sarge setup with samba 3.0.10 with ldap as SAM backend. every thing used to work as expected, but last week the machine went down. an analysis of the problem has showen, that with increasing uptime there is a growing number of smbd processes that caus 100% CPU utilation. tests have showen, that i can kill these processes without harming smb conecctivtiy. this is a workaround but no soulution. i've absolutely no idea what causes this and what i can/should do to track this down. any suggestions ? TIA Matthias Henze matthias@mhcsoftware.de Use PGP!! http://www.mhcsoftware.de/MatthiasHenze.asc - - - - - - - - - - - - - - - - - - - - - - - - - - - MHC SoftWare GmbH voice: +49-(0)9533-92006-0 Fichtera 17 fax: +49-(0)9533-92006-6 96274 Itzgrund/Germany e-Mail: info@mhcsoftware.de - - - - - - - - - - - - - - - - - - - - - - - - - - -
Heya, the growing number of smbd process and therewith the resulting 100% cpu usage have their reason in samba not killing processes which are not used anymore (since every client causes Samba to spawn a new process). You can resolve this issue by putting deadtime = 60 into your smb.conf. This causes Samba to kill processes which are not used anymore after 60 minutes which should be sufficient. Hope I helped! Have a nice day, Ulrik Matthias Henze wrote:> hi, > > i've a debian sarge setup with samba 3.0.10 with ldap as SAM backend. > every thing used to work as expected, but last week the machine went > down. an analysis of the problem has showen, that with increasing uptime > there is a growing number of smbd processes that caus 100% CPU > utilation. tests have showen, that i can kill these processes without > harming smb conecctivtiy. this is a workaround but no soulution. > > i've absolutely no idea what causes this and what i can/should do to > track this down. > > any suggestions ? > > TIA > > > Matthias Henze matthias@mhcsoftware.de > > Use PGP!! http://www.mhcsoftware.de/MatthiasHenze.asc > - - - - - - - - - - - - - - - - - - - - - - - - - - - > MHC SoftWare GmbH voice: +49-(0)9533-92006-0 > Fichtera 17 fax: +49-(0)9533-92006-6 > 96274 Itzgrund/Germany e-Mail: info@mhcsoftware.de > - - - - - - - - - - - - - - - - - - - - - - - - - - -
hi jose, thanks for the hint, but i do not think, dat this causes the problem. i use ldap on the whole system (pam, nss, mail, samba ...) so goups "should" be consistent. # net groupmap list users (S-1-5-21-2136796781-1836600926-1834964671-513) -> users admins (S-1-5-21-2136796781-1836600926-1834964671-512) -> admins guests (S-1-5-21-2136796781-1836600926-1834964671-514) -> guests buchhaltung (S-1-5-21-2136796781-1836600926-1834964671-3005) -> buchhaltung cheers matthias --On Freitag, April 01, 2005 09:27:30 +0200 "Jos? M. Fandi?o" <samba@fadesa.es> wrote:> Matthias Henze wrote: >> >> hi, >> >> thanks for the reply. this solves the problem only partialy. by now, i've >> no more userer porcesses causing this. at the moment i've one root smbd >> that causes 100% system load ... >> >> root 8668 60.8 0.1 10356 2972 ? R Mar31 1173:14 >> /usr/sbin/smbd -D >> >> can some one please help? this is killing me ... > > I remember a thread in the samba mailing list about 100% of cpu use and > the cause was a mapping to an inexistent group or user. > > -- > -----BEGIN GEEK CODE BLOCK----- > Version: 3.1 > GCS/IT d- s+:+() a31 C+++ UBL+++$ P+ L+++ E--- W++ N+ o++ K- w--- > O+ M+ V- PS+ PE+ Y++ PGP+>+++ t+ 5 X+$ R- tv-- b+++ DI D++>+++ > G++ e- h+(++) !r !z > ------END GEEK CODE BLOCK------Matthias Henze matthias@mhcsoftware.de Use PGP!! http://www.mhcsoftware.de/MatthiasHenze.asc - - - - - - - - - - - - - - - - - - - - - - - - - - - MHC SoftWare GmbH voice: +49-(0)9533-92006-0 Fichtera 17 fax: +49-(0)9533-92006-6 96274 Itzgrund/Germany e-Mail: info@mhcsoftware.de - - - - - - - - - - - - - - - - - - - - - - - - - - -
i've killed the process. with the next "bad" process i will do this. this may take some time, as with "deadtime", as proposed by ulrik, i've reduced the numer of such processes dramaticly. but this "deadtime" stuff looks to me like a workaround, no solution, as i do not konw the source of the problem. any way - "deadtime" made it a bit easier for me. my main concern is that too many of these processes crash the machine and i can't monitor it all the time .... --On Freitag, April 01, 2005 10:22:46 +0200 "Jos? M. Fandi?o" <samba@fadesa.es> wrote:> Matthias Henze wrote: >> >> hi jose, >> >> thanks for the hint, but i do not think, dat this causes the problem. i >> use ldap on the whole system (pam, nss, mail, samba ...) so goups >> "should" be consistent. >> >> # net groupmap list >> users (S-1-5-21-2136796781-1836600926-1834964671-513) -> users >> admins (S-1-5-21-2136796781-1836600926-1834964671-512) -> admins >> guests (S-1-5-21-2136796781-1836600926-1834964671-514) -> guests >> buchhaltung (S-1-5-21-2136796781-1836600926-1834964671-3005) -> >> buchhaltung > > > could you run "strace" with the "-p" parameter pointing the proccess > with high cpu use? > ># strace -h > ... > -p pid -- trace process with process id PID, may be repeated > > -- > -----BEGIN GEEK CODE BLOCK----- > Version: 3.1 > GCS/IT d- s+:+() a31 C+++ UBL+++$ P+ L+++ E--- W++ N+ o++ K- w--- > O+ M+ V- PS+ PE+ Y++ PGP+>+++ t+ 5 X+$ R- tv-- b+++ DI D++>+++ > G++ e- h+(++) !r !z > ------END GEEK CODE BLOCK------Matthias Henze matthias@mhcsoftware.de Use PGP!! http://www.mhcsoftware.de/MatthiasHenze.asc - - - - - - - - - - - - - - - - - - - - - - - - - - - MHC SoftWare GmbH voice: +49-(0)9533-92006-0 Fichtera 17 fax: +49-(0)9533-92006-6 96274 Itzgrund/Germany e-Mail: info@mhcsoftware.de - - - - - - - - - - - - - - - - - - - - - - - - - - -