Thread: MemoryContextAlloc: invalid request size
<p><font size="2">Our customer is reporting a database problem after they ran into a </font><br /><font size="2">disk
quota/filesystemfull situation.</font><p><font size="2">This is Postgres 7.2.x on Solaris.</font><br /><font
size="2">Thedatabase server starts without any obvious errors, but the app. Server</font><br /><font size="2">cannot
establishconnection. (It is setup to use Unix Domain Socket)</font><br /><font size="2"> </font><br /><font size="2">I
didnot have a chance to see the database-log, but we have the error-log and </font><br /><font size="2">the truss log
fromthe App. Server.</font><p><font size="2">In the error-log I see this:</font><br /><font size="2"> ...
...</font><br/><font size="2"> Connection to database failed</font><br /><font size="2"> FATAL: MemoryContextAlloc:
invalidrequest size 0</font><br /><p><font size="2">And this in the trace log:</font><br /><font size="2">...
...</font><br/><font size="2">10894: open("/usr/lib/libresolv.so.2", O_RDONLY) = 11</font><br /><font
size="2">10894: fstat(11, 0xFFBE9004) = 0</font><br /><font size="2">10894: mmap(0xFE020000,
8192,PROT_READ|PROT_EXEC, MAP_PRIVATE|MAP_FIXED, 11, 0) = 0xFE020000</font><br /><font size="2">10894:
mmap(0x00000000,303104, PROT_READ|PROT_EXEC, MAP_PRIVATE, 11, 0) = 0xFDD20000</font><br /><font size="2">10894:
mmap(0xFDD64000,15564, PROT_READ|PROT_WRITE|PROT_EXEC, MAP_PRIVATE|MAP_FIXED, 11, 212992) = 0xFDD64000</font><br
/><fontsize="2">10894: mmap(0xFDD68000, 2728, PROT_READ|PROT_WRITE|PROT_EXEC, MAP_PRIVATE|MAP_FIXED|MAP_ANON, -1, 0) =
0xFDD68000</font><br/><font size="2">10894: munmap(0xFDD54000, 65536) = 0</font><br /><font
size="2">10894: memcntl(0xFDD20000, 33536, MC_ADVISE, MADV_WILLNEED, 0, 0) = 0</font><br /><font size="2">10894:
close(11) = 0</font><br /><font size="2">10894: munmap(0xFE020000,
8192) = 0</font><br /><font size="2">10894: so_socket(1, 2, 0, "", 1) =
11</font><br/><font size="2">10894: fstat64(11, 0xFFBE8A70) = 0</font><br /><font
size="2">10894: getsockopt(11, 65535, 8192, 0xFFBE8B70, 0xFFBE8B6C, 0) = 0</font><br /><font size="2">10894:
setsockopt(11,65535, 8192, 0xFFBE8B70, 4, 0) = 0</font><br /><font size="2">10894: fcntl(11, F_SETFL,
0x00000080) = 0</font><br /><font size="2">10894: connect(11, 0x0060EAA0, 77, 1) =
0</font><br/><font size="2">10894: poll(0xFFBE89E8, 1, -1) = 1</font><br /><font
size="2">10894: sigaction(SIGPIPE, 0xFFBE8788, 0xFFBE8888) = 0</font><br /><font size="2">10894: send(11, "\0\001
(\002\0\0s y n c".., 296, 0) = 296</font><br /><font size="2">10894: sigaction(SIGPIPE, 0xFFBE8788, 0xFFBE8888)
=0</font><br /><font size="2">10894: poll(0xFFBE89E8, 1, -1) = 1</font><br /><font
size="2">10894: recv(11, " R\0\0\0\0 E F A T A L :".., 16384, 0) = 58</font><br /><font size="2">... ...</font><p><font
size="2">Couldyou recommend the remedy?</font><p><font size="2">Thanks in advance,</font><br /><font
size="2">Mike.</font>
I got the copy of the database and ran it with truss.
From the trace log it looked like ${PGDATA/}base/<last-number>/pg_internal.init was corrupted
I replaced this file with ${PGDATA/}base/<prev-number>/pg_internal.init
After that I was able to login and use database. Still don’t understand what I’ve done.
Can somebody shed little light on what actually happened?
Thanks,
Mike
From: Brusser, Michael [mailto:Michael.Brusser@matrixone.com]
Sent: Thursday, June 16, 2005 8:28 PM
To: 'Pgsql-Hackers (pgsql-hackers@postgresql.org)'
Subject: [HACKERS] MemoryContextAlloc: invalid request size
Our customer is reporting a database problem after they ran into a
disk quota/filesystem full situation.
This is Postgres 7.2.x on Solaris.
The database server starts without any obvious errors, but the app. Server
cannot establish connection. (It is setup to use Unix Domain Socket)
I did not have a chance to see the database-log, but we have the error-log and
the truss log from the App. Server.
In the error-log I see this:
... ...
Connection to database failed
FATAL: MemoryContextAlloc: invalid request size 0
And this in the trace log:
... ...
10894: open("/usr/lib/libresolv.so.2", O_RDONLY) = 11
10894: fstat(11, 0xFFBE9004) = 0
10894: mmap(0xFE020000, 8192, PROT_READ|PROT_EXEC, MAP_PRIVATE|MAP_FIXED, 11, 0) = 0xFE020000
10894: mmap(0x00000000, 303104, PROT_READ|PROT_EXEC, MAP_PRIVATE, 11, 0) = 0xFDD20000
10894: mmap(0xFDD64000, 15564, PROT_READ|PROT_WRITE|PROT_EXEC, MAP_PRIVATE|MAP_FIXED, 11, 212992) = 0xFDD64000
10894: mmap(0xFDD68000, 2728, PROT_READ|PROT_WRITE|PROT_EXEC, MAP_PRIVATE|MAP_FIXED|MAP_ANON, -1, 0) = 0xFDD68000
10894: munmap(0xFDD54000, 65536) = 0
10894: memcntl(0xFDD20000, 33536, MC_ADVISE, MADV_WILLNEED, 0, 0) = 0
10894: close(11) = 0
10894: munmap(0xFE020000, 8192) = 0
10894: so_socket(1, 2, 0, "", 1) = 11
10894: fstat64(11, 0xFFBE8A70) = 0
10894: getsockopt(11, 65535, 8192, 0xFFBE8B70, 0xFFBE8B6C, 0) = 0
10894: setsockopt(11, 65535, 8192, 0xFFBE8B70, 4, 0) = 0
10894: fcntl(11, F_SETFL, 0x00000080) = 0
10894: connect(11, 0x0060EAA0, 77, 1) = 0
10894: poll(0xFFBE89E8, 1, -1) = 1
10894: sigaction(SIGPIPE, 0xFFBE8788, 0xFFBE8888) = 0
10894: send(11, "\0\001 (\002\0\0 s y n c".., 296, 0) = 296
10894: sigaction(SIGPIPE, 0xFFBE8788, 0xFFBE8888) = 0
10894: poll(0xFFBE89E8, 1, -1) = 1
10894: recv(11, " R\0\0\0\0 E F A T A L :".., 16384, 0) = 58
... ...
Could you recommend the remedy?
Thanks in advance,
Mike.