Mcannally: Maybe time to get back into installing a more recent version of mysql too! and not being dragged back by centos
Bachinski: Robsco: try an alternative yum repo
Deason: Robsco: you can use mysql-community repo from oracle, the mariadb fork’s repo, or something like ius community repo
Deason: Robsco: all three are fine choices.
Deason: Robsco: there is also percona server repo, which is another fork, but a heads up that cent7 will use mariadb by default, so you can get ahead of the curve
Muschik: Thanks both, will look into them all, for something so “straight-forward” as a db, we should really be on something a bit more current
Brass: If I can just get these 12m rows imported tonight I’ll be chuffed, 4.5m so far and all looking goof
Deason: Robsco: ius packages try to look like rhel native packages ala epel but without replacing existing packages, invented before repo.mysql.com
Bousum: A gl*** or two of wine always helps :
Deason: Robsco: repo.mysql.com is oracle, so straight from the source
Zanderigo: Deason, thanks, good to know
Duthie: Seeing a great consistent 30s per 100,000 insert now, couldn’t be happier. it’s not amazing, but at least consistent
Deason: Robsco: cent/rhel also use cfq instead of deadline i/o scheduler, which is really unfortunate cent7 uses deadline like a sane os
Deason: Deadline does a better job of consistent i/o than cfq in my experience
Maxham: We’ve not, and prob won’t, get to CentOS 7 anytime soon, Rackspace still have a problem with them for a start, so gives me a year or so
Emmrich: I’m processing like a man man on a Dell ch***is of 16 blades, and replicating to our DC with Rackspace
Stolinski: Now looking for what other data to process :
Deason: I just meant you can use deadline on el6 but it is not the default until el7 ubuntu always used deadline afaik
Haselhuhn: Got another 2 ch***is full to play with as “old kit”.
Morosow: Any ideas on my collation question anyone? :
Lewry: Deason, ok, thanks for the insight
Nidiffer: BlaDe: testing in mysql is easy see http://gtowey.blogspot.com/2009/10/simple-test-case-techiques.html for hints
Deason: BlaDe: you aren’t using binlog_format=row?
Deason: BlaDe: force the query to be unsafe
Bachinski: BlaDe: “The BINARY and VARBINARY types are similar to CHAR and VARCHAR, except that they contain binary strings rather than nonbinary strings. That is, they contain byte strings rather than character strings. This means that they have no character set, and sorting and comparison are based on the numeric values of the bytes in the values.”
Duhon: Bachinski: gotcha, bummer, would’ve been an easy fix!
Bachinski: BlaDe: no collation need apply :-
Hatherly: Deason: Sorry, I have no idea about that manual entry.
Compiseno: Deason: Sorry, I have no idea about that manual entry.
Deason: BlaDe: it was a joke btw
Deason: BlaDe: you should fix python code instead
Condray: Deason: well in this case, replicating inet6_aton to be ran on the slaves is fine
Sarmento: Deason: I am looking at that, but I don’t really know how / understand the problem well enough
Delanoche: Deason: this line fails https://github.com/noplay/python-mysql-replication/blob/master/pymysqlreplication/event.py#L140
Deason: BlaDe: wait, the statement was already unsafe it is a row event?
Dammen: Deason: and it’s unpacked at: https://github.com/noplay/python-mysql-replication/blob/master/pymysqlreplication/packet.py#L72
Bartnett: Deason: it’s an INSERT
Deason: BlaDe: in the binlog, you see: INSERT .?
Deason: BlaDe: what did you mean by “is there a way I can make it replicate the function”