Обсуждение: Got it...and...
perfect, and error message that tells me what to fix: FATAL 1: btree: BTP_CHAIN flag was expected in radhist_userid (access = bt_read) Whoop! :) Rebuilding that index now... But, still have, and don't know where to begin diagnosing it... FATAL 1: palloc failure: memory exhausted Except, the query appears to be: Apr 11 10:25:23 clio radiusd[13005]: query failed: select uniq_id from radlog where uniq_id='237287829' and stop_time=0 and term_server='isdn-1.trends.ca'; When it failed...radlog being 7k records large... Marc G. Fournier Systems Administrator @ hub.org primary: scrappy@hub.org secondary: scrappy@{freebsd|postgresql}.org
The Hermit Hacker wrote: > > perfect, and error message that tells me what to fix: > > FATAL 1: btree: BTP_CHAIN flag was expected in radhist_userid (access = > bt_read) > > Whoop! :) Rebuilding that index now... > > But, still have, and don't know where to begin diagnosing it... > > FATAL 1: palloc failure: memory exhausted > > Except, the query appears to be: > > Apr 11 10:25:23 clio radiusd[13005]: query failed: select uniq_id > from radlog where uniq_id='237287829' and stop_time=0 > and term_server='isdn-1.trends.ca'; > > When it failed...radlog being 7k records large... Did you restart postmaster after FATALs ? Vadim
On Mon, 13 Apr 1998, Vadim B. Mikheev wrote: > The Hermit Hacker wrote: > > > > perfect, and error message that tells me what to fix: > > > > FATAL 1: btree: BTP_CHAIN flag was expected in radhist_userid (access = > > bt_read) > > > > Whoop! :) Rebuilding that index now... > > > > But, still have, and don't know where to begin diagnosing it... > > > > FATAL 1: palloc failure: memory exhausted > > > > Except, the query appears to be: > > > > Apr 11 10:25:23 clio radiusd[13005]: query failed: select uniq_id > > from radlog where uniq_id='237287829' and stop_time=0 > > and term_server='isdn-1.trends.ca'; > > > > When it failed...radlog being 7k records large... > > Did you restart postmaster after FATALs ? This is a production server, so restarting it isn't much of an option...as well, by the time its noticed (several hours after it happens), its too late anyway, no? Marc G. Fournier Systems Administrator @ hub.org primary: scrappy@hub.org secondary: scrappy@{freebsd|postgresql}.org