#maria Log v0.1

logs Catalogue

This page loads from the db, refresh as and when. It displays 24 hrs
For older data http://marialog.archivist.info/previous.php, logs go back to 2009-01-18
Some user stats
Table logs
date_time
user
message
2017-11-22 12:50:55
welker
timestamy
2017-11-22 12:50:57
welker
timestamp
2017-11-22 12:53:51
jkavalik
welker, yes, so do not group by Name :) Name is what you are counting and if you add it to the grouping expression, the count will always be 1
2017-11-22 14:59:26
svar
lifeboy away: innodb_log_file_size and log buffer size should both be increase for insert performance, server io capacity and type of disk is needed for deeper fine tuning ,but in any case as vld mention if ODBC doing single insert it may as well improve nothing
2017-11-22 15:03:48
vvaintroub
svar, yep, first transactions, multi-inserts, then disable_keys/enable_keys, then innodb_flush_log_at_trx_commit=2, and so on
2017-11-22 15:03:54
vvaintroub
but first - transactions :)
2017-11-22 15:37:14
tom[]
fyi, netdata now has some charts for galera https://monosnap.com/file/22HCmPOvhOvqzOzQ6xDtss1bktqGy6.png
2017-11-22 15:40:31
tom[]
i've relied on jayjanssen/myq_gadgets for a long time. this will supplement that
2017-11-22 16:16:58
MadMerlin|work
is wsrep_apply_oooe a value between 0 and 1?
2017-11-22 16:17:14
MadMerlin|work
the docs don't say explicitly (although somewhat imply it)
2017-11-22 16:56:50
tom[]
MadMerlin|work: idk for sure but it looks that way https://github.com/codership/galera/blob/release_25.3.22/galera/src/monitor.hpp
2017-11-22 18:59:15
tom[]
i have to run an UPDATE that changes the value of one column on 40M rows. it takes about 2 hours. would it be quicker to drop the index on that column before the UPDATE and add it back later?
2017-11-22 19:00:59
MadMerlin|work
it's probably worth trying
2017-11-22 19:02:41
bratchley
Where can I find a description of what exactly "--wsrep-new-cluster" does? I mean I know I pass it when starting a new cluster but I'd like to know what I'm actually telling it to do.
2017-11-22 19:03:02
bratchley
I've tried google but all I get are clustering tutorials
2017-11-22 19:03:42
MadMerlin|work
basically, take what's currently in the data dir and treat that as authoritative for what the cluster should contain
2017-11-22 19:03:49
MadMerlin|work
then bootstrap the cluster
2017-11-22 19:04:51
tom[]
and start over with cluster configuration serial number, which is a clue to what "cluster" means in galera context
2017-11-22 19:05:10
bratchley
MadMerlin|work: so nothing gets reset?
2017-11-22 19:05:23
bratchley
reset/overwritten
2017-11-22 19:05:34
tom[]
no data is reset or overwritten
2017-11-22 19:05:41
bratchley
cool beans
2017-11-22 19:05:55
bratchley
ty both
2017-11-22 19:06:04
tom[]
"cluster" in galera is a specific instance of a configuration of nodes. and it changes automatically as nodes join and leave
2017-11-22 19:06:50
tom[]
when a node starts and it cannot find any other nodes to connect to, it refuses to go on.
2017-11-22 19:07:48
tom[]
except when you specify wsrep-new-cluster. in that case, you give that node authority to be the first node in a new cluster that's borne out of nothing
2017-11-22 20:32:44
jak2000
installed new mariadbserver
2017-11-22 20:33:01
jak2000
how to chnge default port, not see in: /etc/mysql/my.cnf
2017-11-22 20:33:39
jak2000
*change
2017-11-22 20:36:04
mrbrklyn
how do I force mariabd to run as a user or group?
2017-11-22 20:39:00
grknight
mrbrklyn: put it in a configuration file or start the daemon with such options
2017-11-22 20:39:18
mrbrklyn
which config file?
2017-11-22 20:39:40
mrbrklyn
my.cnf
2017-11-22 20:39:41
mrbrklyn
?
2017-11-22 20:40:29
grknight
anyone just tell the daemon about it. eg. 'mysqld --defaults-extra-file=myuser.cnf'
2017-11-22 20:40:54
grknight
or --defaults-file to dump the system settings
2017-11-22 20:42:22
mrbrklyn
possibly in mysqld_safe
2017-11-22 20:44:22
grknight
i don't like mysqld_safe because of all the assumptions. if you are starting a server, start it
2017-11-22 20:45:21
mrbrklyn
its what monty taught me to do ;)
2017-11-22 20:47:15
grknight
all I'll say is that it is a script and had a CVE on it at one point (from MySQL) so what was _safe really wasn't
2017-11-22 20:50:39
mrbrklyn
no --defaults-file
2017-11-22 20:54:17
mrbrklyn
[www3 ~]# mysqld --print-defaults
2017-11-22 20:54:17
mrbrklyn
mysqld would have been started with the following arguments:
2017-11-22 20:54:17
mrbrklyn
--skip-networking --port=3306 --socket=/run/mariadb/mysqld.sock --datadir=/usr/local/var --skip-external-locking --key_buffer_size=16M --max_allowed_packet=1M --table_open_cache=64 --sort_buffer_size=512K --net_buffer_length=8K --read_buffer_size=256K --read_rnd_buffer_size=512K --myisam_sort_buffer_size=8M --pid-file=/run/mariadb/mariadb.pid --log-bin=mysql-bin --binlog_format=mixed --server-id=1
2017-11-22 20:57:11
grknight
if you use --defaults-file, you need to give it a file path
2017-11-22 20:57:39
grknight
eg. 'mysqld --defaults-file=exclusive-user.cnf'
2017-11-22 21:02:17
mrbrklyn
I can't get the package manager to stop making the user mysql
2017-11-22 21:02:28
mrbrklyn
in /etc/passwd
2017-11-22 21:06:02
grknight
you'd have to rewrite the package for that
2017-11-22 21:18:59
tmkt
Hey Hey, have 3 mariadb boxes, clustered together (dev env)
2017-11-22 21:19:10
tmkt
power went out, and now they are all hung
2017-11-22 21:20:00
tmkt
i've tried --bootstrap, and various other cli args
2017-11-22 21:20:02
tmkt
without success
2017-11-22 21:20:46
tmkt
Nov 22 16:05:01 dev3 mysqld[5576]: 2017-11-22 16:05:01 140257769662656 [Note] /usr/sbin/mysqld (mysqld 10.2.10-MariaDB-10.2.10+maria~xenial-log) starting as process 5576 .. all of them, that is there last line in the syslog
2017-11-22 21:20:56
tmkt
and 3306 isn't responding...not sure if i just need to wait it out?
2017-11-22 21:21:01
tmkt
its been 30 minutes so far
2017-11-22 21:26:00
tmkt
nodes seem to be talking to each other, but won't talk to me 2017-11-22 16:05:10 140257662654208 [Note] WSREP: New cluster view: global state: eb856dc4-8dc3-11e7-95b8-369d13e17f80:9980591, view# -1: non-Primary, number of nodes: 3, my index: 2, protocol version -1
2017-11-22 21:30:35
dragonheart
Was only one node started with bootstrap? What does show global status like 'wsrep%'; show on all nodes?
2017-11-22 21:31:22
dragonheart
you cannot connect to a database instance until the node is back to the ready status however you should be able to do the above.
2017-11-22 21:31:42
dragonheart
database instance - specific database.
2017-11-22 21:31:55
dragonheart
what sst mechanism do you have?
2017-11-22 21:33:23
tmkt
i can't mysql
2017-11-22 21:34:04
tmkt
I ususally start up the main box with service mysql restart --wsrep-new-cluster
2017-11-22 21:34:15
tmkt
then the other 2 service mysql start and all good
2017-11-22 21:34:46
tmkt
wsrep_sst_method=rsync
2017-11-22 21:35:44
dragonheart
ok - that's good. so you see rsync processes running?
2017-11-22 21:36:27
tmkt
none of the machines have any rsync running
2017-11-22 21:40:06
tmkt
i think i found something
2017-11-22 21:40:06
tmkt
WSREP: It may not be safe to bootstrap the cluster from this node. It was not the last one to leave the cluster and may not contain all the updates. To force cluster bootstrap with this node, edit the grastate.dat file manually and set safe_to_bootstrap to
2017-11-22 21:42:30
MadMerlin|work
they all died at the same time, so you need to pick one to do a new bootstrap from and then do it
2017-11-22 21:42:43
MadMerlin|work
(which likely involves following the above note)
2017-11-22 21:43:27
tmkt
yeah...so grabbed the original one
2017-11-22 21:44:12
tmkt
much better, see rsync processes now
2017-11-22 21:45:20
tmkt
main one is responding now, just have to wait it out now?
2017-11-22 21:46:00
MadMerlin|work
if there's sst involved (which there probably is) then the donors won't be responsive until that's done
2017-11-22 21:47:13
tmkt
ok...makes sense..
2017-11-22 21:47:18
tmkt
almost 5 anyhow
2017-11-22 21:47:26
tmkt
should be good tomorrow morning ideally
2017-11-22 21:50:18
tmkt
Thanks!
2017-11-23 04:00:43
Bhakimi
hi guys, is there a way to make mariadb store the master status info in a file so when mysql is down i can still get that data?
2017-11-23 04:14:20
dragonheart
Bhakimi: you are probably able to start it with skip-networking or in a readonly mode to access the status however.
2017-11-23 04:15:18
Bhakimi
yea i figured
2017-11-23 04:15:33
Bhakimi
but i rememver in the past it used to store it somehow in a master.info file
2017-11-23 04:16:22
dragonheart
it was subject to race conditions compared to the actual commit. hence the change.
2017-11-23 04:16:41
Bhakimi
something likr --master-info-file= but for the master
2017-11-23 04:17:17
Bhakimi
i see
2017-11-23 04:17:31
Bhakimi
k i'll start it with skip-netwoprk and connect to it via socket
2017-11-23 04:18:11
dragonheart
which version are you running?
2017-11-23 04:28:43
gh0st3d
Hey everyone... trying to figure out a good database design. We have users (nothing more than an ID) who have profiles (many fields, varying data types, list of fields may grow/change over time)... Two of the fields require a full TEXT value while the rest could be saved as VARCHAR with limited space. How should I go about doing this efficiently?
2017-11-23 04:32:57
dragonheart
gh0st3d: one unsigned (big?) int for the id, a text field for the JSON, and virtual columns and index on any JSON field you need indexed - https://mariadb.org/wp-content/uploads/2017/11/JSON-MariaDB.pdf
2017-11-23 04:33:47
gh0st3d
hmm so you're saying 1 table for all of it and just cramming the whole profile into one JSON string?
2017-11-23 04:35:18
dragonheart
yes. haven't seen a compelling reason to go beyond that at this stage. how you select/join/use types might influence that however.
2017-11-23 04:37:34
gh0st3d
Gotcha, I may just have a case of trying to over optimize before it's necessary. Plan on setting it up with an API, huge majority of the time would just be pulling the whole profile. Probably add a field for "last updated" so I can cache responses & bust the cache if it's been updated since
2017-11-23 04:37:38
gh0st3d
Thanks!
2017-11-23 04:38:57
dragonheart
frequent updates might put that as a separate index column but otherwise, start basic and see how you go.
2017-11-23 04:39:53
dragonheart
of course a KV store would do this quite well without a database functionality.
2017-11-23 07:40:35
salle
.