Details
-
Bug
-
Status: Closed (View Workflow)
-
Major
-
Resolution: Incomplete
-
10.1.21
-
CloudLinux Server release 6.9
mysql Ver 15.1 Distrib 10.1.21-MariaDB, for Linux (x86_64) using readline 5.1
kernel: 2.6.32-673.8.1.lve1.4.3.1.el6.x86_64
Description
MariaDB suddenly crashes with [ERROR] mysqld got signal 11 ; during the request:
*************************** 8. row ***************************
|
Id: 121063
|
User: comua120_usr
|
Host: localhost
|
db: comua120_back_09-03-2017
|
Command: Query
|
Time: 5
|
State: statistics
|
Info: SELECT fl.name feature_name, fp.id_feature, fv.id_feature_value, fvl.value,
|
COUNT(DISTINCT p.id_product) nbr,
|
lifl.url_name name_url_name, lifl.meta_title name_meta_title, lifvl.url_name value_url_name, lifvl.meta_title value_meta_title , psi.price_min, psi.price_max
|
|
FROM ps_feature_product fp
|
INNER JOIN ps_product p ON (p.id_product = fp.id_product)
|
LEFT JOIN ps_feature_lang fl ON (fl.id_feature = fp.id_feature AND fl.id_lang = 2)
|
INNER JOIN ps_feature_value fv ON (fv.id_feature_value = fp.id_feature_value AND (fv.custom IS NULL OR fv.custom = 0))
|
LEFT JOIN ps_feature_value_lang fvl ON (fvl.id_feature_value = fp.id_feature_value AND fvl.id_lang = 2)
|
LEFT JOIN ps_layered_indexable_feature_lang_value lifl
|
ON (lifl.id_feature = fp.id_feature AND lifl.id_lang = 2)
|
LEFT JOIN ps_layered_indexable_feature_value_lang_value lifvl
|
ON (lifvl.id_feature_value = fp.id_feature_value AND lifvl.id_lang = 2) INNER JOIN ps_product_shop product_shop
|
ON (product_shop.id_product = p.id_product AND product_shop.id_shop = 1)
|
|
INNER JOIN `ps_layered_price_index` psi ON (psi.id_product = p.id_product AND psi.id_currency = 1
|
AND psi.price_min <= 16000 AND psi.price_max >= 0)
|
WHERE product_shop.`active` = 1 AND fp.id_feature = 9
|
AND p.id_product IN (
|
SELECT id_product
|
FROM ps_category_product cp
|
INNER JOIN ps_category c ON (c.id_category = cp.id_category AND
|
c.nleft >= 25
|
AND c.nright <= 50
|
AND c.active = 1)) AND p.id_product IN (SELECT id_product FROM ps_feature_product fp WHERE fp.`id_feature_value` = 92) AND p.id_product IN (SELECT id_product FROM ps_feature_product fp WHERE fp.`id_feature_value` = 37) AND p.id_product IN (SELECT id_product FROM ps_feature_product fp WHERE fp.`id_feature_value` = 31) AND p.id_product IN (SELECT id_product FROM ps_feature_product fp WHERE fp.`id_feature_value` = 59)
|
GROUP BY fv.id_feature_value
|
Progress: 0.000
|
As a workaround we had to dump and restore database into the new db. This process (dump/restore) should be made every 10-14 days, as in the other case we are facing this bug again.
my.cnf:
[mysqld]
|
max_connections = 300
|
max_user_connections = 30
|
|
innodb_file_per_table = 1
|
innodb_corrupt_table_action = assert
|
local-infile = 0
|
|
key_buffer_size = 512M
|
query_cache_size = 96M
|
query_cache_limit = 1M
|
query_cache_min_res_unit = 2048
|
|
table_open_cache = 7000
|
innodb_open_files = 7000
|
open_files_limit=50000
|
|
tmp_table_size = 16M
|
max_heap_table_size = 16M
|
|
join_buffer_size = 32M
|
max_join_size = 100000000
|
|
thread_cache_size = 20
|
|
innodb_buffer_pool_size = 8000M
|
innodb_buffer_pool_instances = 8
|
innodb_flush_log_at_trx_commit = 2
|
innodb_log_buffer_size = 16M
|
innodb_additional_mem_pool_size = 24M
|
innodb_log_file_size = 512M
|
innodb_io_capacity = 200
|
|
max_allowed_packet = 268435456
|
|
tmpdir=/var/lib/mysql/tmp_table
|