Re: 3.23.42 first crash (fwd) - Martin MOKREJŠ - com.mysql.lists.bugs

10 messages in com.mysql.lists.bugsRe: 3.23.42 first crash (fwd)

From	Sent On	Attachments
Martin MOKREJŠ	11 Sep 2001 08:11
Michael Widenius	12 Sep 2001 04:25
Martin MOKREJŠ	12 Sep 2001 06:09
Martin MOKREJŠ	17 Sep 2001 05:34
Sasha Pachev	17 Sep 2001 06:23
Martin MOKREJŠ	17 Sep 2001 07:35
Michael Widenius	18 Sep 2001 04:36
Martin MOKREJŠ	18 Sep 2001 09:11
Michael Widenius	18 Sep 2001 10:47
Martin MOKREJŠ	19 Sep 2001 07:25

Subject:	Re: 3.23.42 first crash (fwd)
From:	Martin MOKREJŠ (mmok...@natur.cuni.cz)
Date:	09/17/2001 05:34:06 AM
List:	com.mysql.lists.bugs

Hi all, I have some more details:

on Wed, 12 Sep 2001, Michael Widenius wrote:

mmokrejs> Hi, I'm running your linux binaries on Linux 2.4.9 with some patch to
fix memory allocation problems (seems it helped). Today I found that mysqld
crashed. Here's the resolved stack:

mmokrejs> 0x807b72f handle_segfault__Fi + 383 mmokrejs> 0x812b92a pthread_sighandler + 154 mmokrejs> 0x81536d7 memcpy + 39 mmokrejs> 0x8077937 net_store_data__FP6StringPCcUi + 155 mmokrejs> 0x8068b14 send__5FieldP6String + 196 mmokrejs> 0x804c153 send__10Item_fieldP6String + 19 mmokrejs> 0x8073cbc send_data__11select_sendRt4List1Z4Item + 92 mmokrejs> 0x80a596c end_send__FP4JOINP13st_join_tableb + 96 mmokrejs> 0x80a0cc8 sub_select__FP4JOINP13st_join_tableb + 200 mmokrejs> 0x80a0a43 do_select__FP4JOINPt4List1Z4ItemP8st_tableP9Procedure + 387 mmokrejs> 0x8099f70
mysql_select__FP3THDP13st_table_listRt4List1Z4ItemP4ItemRt4List1Z15Item_func_matchP8st_orderT5P4ItemT5UiP13select_result
+ 5600 mmokrejs> 0x8081e6f mysql_execute_command__Fv + 707 mmokrejs> 0x8085c96 mysql_parse__FP3THDPcUi + 210 mmokrejs> 0x80812bd do_command__FP3THD + 1261 mmokrejs> 0x80807bc handle_one_connection__FPv + 548

mmokrejs> key_buffer_size=268431360 mmokrejs> record_buffer=67104768 mmokrejs> sort_buffer=67108856 mmokrejs> max_used_connections=0 mmokrejs> max_connections=100 mmokrejs> threads_connected=1

mmokrejs> I'll turn on the logging facility so that we know what was the sql
command.

Description:

So, finally during weekend it crashed again.

0x807b72f handle_segfault__Fi + 383 0x812b92a pthread_sighandler + 154 0x81536e3 memcpy + 51 0x80fbd3e _mi_pack_get_block_info + 394 0x80fc0da _mi_read_pack_record + 54 0x80f46e5 mi_rkey + 453 0x80c6209 index_read_idx__9ha_myisamPcUiPCcUi16ha_rkey_function + 53 0x80a130b join_read_const__FP13st_join_table + 95 0x80a1108 join_read_const_tables__FP4JOIN + 132 0x8099162
mysql_select__FP3THDP13st_table_listRt4List1Z4ItemP4ItemRt4List1Z15Item_func_matchP8st_orderT5P4ItemT5UiP13select_result
+ 2002 0x8081e6f mysql_execute_command__Fv + 707 0x8085c96 mysql_parse__FP3THDPcUi + 210 0x80812bd do_command__FP3THD + 1261 0x80807bc handle_one_connection__FPv + 548

Some pointers may be invalid and cause the dump to abort... thd->query at 0x5a187ec0 is invalid pointer thd->thread_id=6911 [...] 010917 11:05:44 mysqld restarted

6911 Connect pedant@localhost on Athaliana_I 6911 Query show tables 6911 Query show tables 6911 Quit

I don't remember ever seeing a bug in this code. Could you try to find a way to repeat the above; It would be very interesting to know what would cause this.

I think it may be related to these errors found in errorrlog:

010910 18:13:52 read_const: Got error 127 when reading table
./Ncrassa/contig_data 010910 18:56:21 read_const: Got error 127 when reading table
./Hpylori99/blast_data 010910 21:44:57 read_const: Got error 127 when reading table
./Halobacterium_NRC_1_pNRC100/blastindex 010911 10:08:04 read_const: Got error 127 when reading table
./Athaliana_II/contig_data 010911 11:09:59 read_const: Got error 127 when reading table
./Athaliana_II/contig_data 010911 12:17:13 read_const: Got error 127 when reading table
./Phorikoshii/blast_data 010911 14:53:17 read_const: Got error 127 when reading table
./Athaliana_II/contig_data 010911 15:03:03 read_const: Got error 127 when reading table
./Athaliana_II/contig_data 010911 19:32:37 read_const: Got error 127 when reading table
./Bacteroides_fragillis_NCTC9343/contig_data 010911 20:33:05 read_const: Got error 127 when reading table
./Athaliana_I/scop2_data 010911 20:33:13 read_const: Got error 127 when reading table
./Athaliana_I/scop1_data 010912 0:08:44 read_const: Got error 127 when reading table
./Athaliana_I/blast_Yeast_data 010912 0:11:36 read_const: Got error 127 when reading table
./Athaliana_I/blast_Human_data 010912 0:12:35 read_const: Got error 127 when reading table
./Athaliana_I/blast_Celegans_data 010912 13:21:00 read_const: Got error 127 when reading table
./Athaliana_I/contig_data 010912 14:46:42 read_const: Got error 127 when reading table
./Athaliana_II/contig_data 010912 15:10:15 read_const: Got error 127 when reading table
./Pasteurella_multocida_PM70/blast_data 010912 17:54:00 read_const: Got error 127 when reading table
./Athaliana_II/contig_data 010913 1:02:01 read_const: Got error 127 when reading table
./Bacteroides_fragillis_NCTC9343/blast_data 010913 10:24:49 read_const: Got error 127 when reading table
./Facidarmanus/contig_data 010913 10:49:58 read_const: Got error 127 when reading table
./Anabaena/contig_data 010913 11:45:27 read_const: Got error 127 when reading table
./Bcepacia/contig_data 010913 12:32:22 read_const: Got error 127 when reading table
./Rsphaeroides/contig_data 010914 14:00:33 read_const: Got error 127 when reading table
./Athaliana_III/contig_data 010914 15:35:56 read_const: Got error 127 when reading table
./Athaliana_II/contig_data 010916 9:34:37 read_const: Got error 127 when reading table
./Facidarmanus/contig_data

However, it looks when mysqld crashed there's not this errormessage printed into
errorlog.

All of these tables are Compressed tables, if it helps.

It looks like somehow a field object pointed to a wrong memory location.

The problem is probably that you did run out of memory when handling blocs becasue your buffers are way too big.

I must say that the kernel 2.4.9+patch does not complain anymore about inability
to allocate memory.