| === Lcawte is now known as Lcawte|Away | ||
| mapps | hmmmmmmmmmmmm] | 07:36 |
|---|---|---|
| MartijnVdS | http://www.estofex.org/cgi-bin/polygon/showforecast.cgi?text=yes&fcstfile=2014060806_201406062211_1_stormforecast.xml | 08:49 |
| brobostigon | morning boys and girls. | 09:29 |
| foobarry | finished my massive lancaster in time for d-day | 09:56 |
| foobarry | http://i.imgur.com/YzSsD5e.jpg | 09:56 |
| foobarry | http://i.imgur.com/D16flk6.jpg | 09:59 |
| bigcalm | Good morning peeps :) | 10:07 |
| === ikonia_ is now known as ikonia | ||
| NET||abuse | quick question, i have an external disk, what's the fastest fs to run for a backup, offloading my backup files from laptop, about 100GB worth, and updating :) | 14:02 |
| NET||abuse | so i want to put ext4 on the disk, but there are options, eg, btrf, xfs, fat32... just being informed from the dropdown on gparted | 14:03 |
| daftykins | NET||abuse: "KISS" | 15:13 |
| === iceflatline is now known as Guest40453 | ||
| foobarry | hilarious bing advert thread on facebook | 15:52 |
| foobarry | read the comments https://www.facebook.com/BingUK/posts/842486845766093 | 15:53 |
| daftykins | there aren't any | 15:54 |
| daftykins | requires login to see perhaps? | 15:54 |
| foobarry | mebbe | 16:05 |
| bigcalm | popey: ping | 16:59 |
| popey | yo | 17:09 |
| bigcalm | Argh | 19:32 |
| penguin42 | argh? | 19:32 |
| popey | ARGH! | 19:33 |
| bigcalm | df shows that /tmp is at 100% usage with 1.9GB | 19:33 |
| bigcalm | du -chs /tmp shows 136K | 19:33 |
| bigcalm | This is on a client's server | 19:33 |
| bigcalm | Where /tmp is its own mount | 19:33 |
| ali1234 | foobarry: the really hilarious thing is when you go to bingiton.com and click "learn about the study" at the very bottom of the page | 19:33 |
| penguin42 | bigcalm: Is it a real /tmp or is it a shmfs? | 19:33 |
| penguin42 | bigcalm: One thing that causes that is a deleted file that's still open | 19:33 |
| bigcalm | From mount: /dev/mapper/vg_root-lv_tmp on /tmp type ext3 (rw) | 19:34 |
| bigcalm | I see | 19:34 |
| penguin42 | nod, ok, so probably a deleted file that's still open | 19:34 |
| bigcalm | How would I find such a file? :) | 19:34 |
| penguin42 | bigcalm: You could try lsof | grep deleted | 19:35 |
| penguin42 | bigcalm: Deleted files don't free the space until every user closes them | 19:35 |
| bigcalm | Ta | 19:36 |
| bigcalm | Wow, that's a lot of open files | 19:37 |
| bigcalm | All of them by mysql :) | 19:38 |
| bigcalm | Screw it, I'm going to stop mysql and see what breaks | 19:40 |
| bigcalm | That's what weekends are for, right? | 19:40 |
| penguin42 | bigcalm: but you greped for deleted ? | 19:40 |
| bigcalm | penguin42: yes | 19:41 |
| bigcalm | penguin42: http://discworld.cuth.eu/dump/deleted.txt | 19:44 |
| penguin42 | wth is cuth? | 19:45 |
| bigcalm | It's me :) | 19:45 |
| penguin42 | bigcalm: mysqld 3353 mysql 253u REG 253,2 1902923776 23 /tmp/ML9IzY9V (deleted) | 19:45 |
| bigcalm | Yep | 19:45 |
| moreati | bigcalm: mario? | 19:45 |
| bigcalm | Heh, hiya moreati :) | 19:46 |
| bigcalm | moreati: how was the barcamp? | 19:46 |
| penguin42 | bigcalm: You got some type of dump/backup script that does crazy things like that? | 19:46 |
| moreati | good thanks | 19:46 |
| ali1234 | you can view the file contents through /proc | 19:46 |
| ali1234 | if you haven't already restarted it | 19:46 |
| bigcalm | I know what the content is. I'm trying to make things usable again | 19:47 |
| bigcalm | If I stop mysql, I'm guessing that the /tmp files will become visible or actually get deleted? | 19:47 |
| ali1234 | they get deleted yes | 19:47 |
| ali1234 | all you really need to do is make mysql close that file | 19:47 |
| bigcalm | Know how? | 19:48 |
| ali1234 | well, what's in the file? | 19:48 |
| bigcalm | Oh, um | 19:48 |
| bigcalm | How do I "view the file contents through /proc"? :) | 19:48 |
| ali1234 | look in /proc/3353/fd with root | 19:49 |
| ali1234 | that's all the file descriptors opened by mysql, by fd number | 19:49 |
| ali1234 | it will be one of those | 19:49 |
| bigcalm | Ta | 19:49 |
| ali1234 | possibly 23? not sure how to read the output of lsof | 19:49 |
| ali1234 | yeah 23 | 19:51 |
| bigcalm | /proc/23 doesn't exist | 19:52 |
| ali1234 | /proc/3353/fd/23 | 19:52 |
| penguin42 | bigcalm: do an ls -l /proc/3353/fd it'll show you what fd is | 19:52 |
| penguin42 | bigcalm: for pid 3353 | 19:52 |
| bigcalm | I seriously don't understand what I'm doing now :) | 19:55 |
| ali1234 | it's probably a temporary file for a query | 19:56 |
| ali1234 | try 'file /proc/3353/fd/23' | 19:57 |
| ali1234 | or hexdump -C | 19:57 |
| bigcalm | ali1234: that would make sense as things broke when I tried to run a query to delete a lot of data from a table | 19:57 |
| ali1234 | did you do a massive join? | 19:57 |
| bigcalm | # file /proc/3353/fd/23 | 19:57 |
| bigcalm | /proc/3353/fd/23: symbolic link to `/data02/mysql/mysql/tables_priv.MYI' | 19:57 |
| bigcalm | No, I did a massive delete | 19:58 |
| ali1234 | hmm that's the wrong fd then | 19:58 |
| bigcalm | The table in question is currently over 15GB | 19:58 |
| bigcalm | I'm trying to delete some of it so that the server's hdd doesn't get full and break the client's website | 19:58 |
| ali1234 | what query did you do? | 19:59 |
| bigcalm | DELETE FROM tb_mass_email_recipient WHERE enu_status = "sent" AND dat_sent < @archive_date; | 20:00 |
| bigcalm | Oh, and this before it: | 20:00 |
| bigcalm | SET @archive_date = DATE_SUB(NOW(), INTERVAL 30 DAY); | 20:00 |
| ali1234 | hmm... | 20:00 |
| ali1234 | well, most likely you'll have to restart mysql | 20:01 |
| bigcalm | I need to sort out the server so that I can start again and set the interval to be 365 days and slowly work towards now | 20:01 |
| ali1234 | is that query still running? | 20:01 |
| bigcalm | No, it broke after 2 hours | 20:01 |
| ali1234 | ah... well, that's definitely the problem then | 20:01 |
| bigcalm | And now I can't view any table because /tmp is full | 20:01 |
| bigcalm | Here goes | 20:02 |
| ali1234 | restart it, and try running the query with a limit clause | 20:02 |
| ali1234 | i dunno if that will help, but it might | 20:02 |
| ali1234 | i mean maybe it's trying to do the delete atomically | 20:02 |
| bigcalm | I've just issued stop, it's taking a while | 20:03 |
| bigcalm | Yay, /tmp is showing 1.8GB free now | 20:03 |
| bigcalm | There are 2.2 million rows in this table I'm trying to archive | 20:04 |
| bigcalm | Great, crashed tables. Why am I not surprised? | 20:08 |
| ali1234 | well yeah. | 20:08 |
| bigcalm | I may regret having issued REPAIR TABLE from my SQL client | 20:09 |
| bigcalm | Now I know to keep an eye on /tmp | 20:11 |
| bigcalm | Whom ever designed this one table - I wish them harm | 20:18 |
| ali1234 | caching outgoing email into a table seems like a common design pattern | 20:23 |
| ali1234 | amusing when someone moves the website to another server and doesn't realise it requires a cronjob to actually send the email | 20:24 |
| bigcalm | But wasn't a job to clear out the cached data once the emails had been sent | 20:30 |
| ali1234 | yeah that's low priority :( | 20:33 |
| bigcalm | Crikey, I'm having to run the query at most 10 days at a time. Might even have to drop it to 1 day per query | 21:38 |
| bigcalm | Well, query set | 21:38 |
| bigcalm | I'm also running OPTIMISE TABLE after the delete to free up space | 21:38 |
| Armag3dd0n24 | what is the best mobile network for free Internet? | 21:44 |
| Armag3dd0n24 | obviously in the UK lol | 21:44 |
| ali1234 | bigcalm: maybe try DELETE QUICK, see http://dev.mysql.com/doc/refman/5.0/en/delete.html | 21:44 |
| ali1234 | or add a limit clause | 21:44 |
| bigcalm | ali1234: will give it a go, ta | 21:45 |
| ali1234 | "If you are going to delete many rows from a table, it might be faster to use DELETE QUICK followed by OPTIMIZE TABLE. This rebuilds the index rather than performing many index block merge operations." | 21:46 |
| bigcalm | Ta | 21:46 |
| bigcalm | Though it's also about freeing up space on the HDD | 21:46 |
| ali1234 | yeah, the optimize should do that | 21:46 |
| bigcalm | Ah, still use the optimize then, goodo | 21:47 |
| ali1234 | if you just delete a row it would just leave a gap in the table file | 21:47 |
| ali1234 | so it has to remove that gap... that's probably the bottleneck if you just use regular delete | 21:48 |
| bigcalm | I have a feeling I should run optimize on some other client's tables :) | 21:48 |
| bigcalm | I see | 21:48 |
| ali1234 | i'm uncompressing 120000 tar.bz2 files, each one is about 1MB and contains around 50-100 files... the whole thing is on a XFS filesystem using deadline schedule... guess how long it will take? | 21:50 |
| bigcalm | A while? | 21:51 |
| penguin42 | so 120GB of data uncompressed or compressed? | 21:51 |
| penguin42 | so ok 120GB of data but dominated by small files - which ain't XFSs strong point mind | 21:52 |
| penguin42 | ali1234: I'll toss my bet in at 15mins | 21:52 |
| ali1234 | yeah, and deadline optimizes for reading... and it's a 5400 rpm drive | 21:53 |
| penguin42 | ohhhhh | 21:53 |
| ali1234 | it's been going for over a day and it's done about 30% | 21:53 |
| penguin42 | ouch | 21:53 |
| ali1234 | i changed it to cfq and it's going much faster now | 21:54 |
| penguin42 | wow that big a difference? | 21:54 |
| ali1234 | seems about 4x faster | 21:54 |
| ali1234 | still freezes up the computer every now and then | 21:55 |
| penguin42 | probably when it realises it's filled RAM up with stuff to be written | 21:55 |
| ali1234 | yeah... i've got about 8GB in the disk cache and 1GB used on swap | 21:56 |
| ali1234 | once i've decompressed it all then i'm going to run findup -m on it and recompress it into larger archives, this should give a 10x decrease in compressed size if my experiments on a small sample are correct | 21:58 |
| ali1234 | how do i check the free inode count on xfs? | 22:00 |
| penguin42 | xfs is always a bit of a mystery to me | 22:00 |
| ali1234 | df -i apparently | 22:00 |
| ali1234 | i forgot this drive was xfs... | 22:00 |
| ali1234 | should have reformatted it before i started | 22:00 |
| penguin42 | oh df -i is just the standard | 22:01 |
| ali1234 | hmm. only used 8% of inodes... should be okay | 22:01 |
| bigcalm | I'm considering setting up a rack in my garage once I move house. That said, I don't know where to start in sourcing parts. Should I just buy something from HP/Dell/eBay or build something from scratch? | 22:10 |
| ali1234 | you mean the actual rack itself? | 22:10 |
| ali1234 | you can build a rack fairly easily, you just need some L-bar and some DIY skills | 22:11 |
| bigcalm | Both rack and server to go in it | 22:13 |
| ali1234 | it depends how fancy you want it really | 22:13 |
| bigcalm | I need a powerful set-up for running multiple VMs at the same time. The HP Proliant microserver I have isn't up to the job. | 22:14 |
| bigcalm | It'll basically take over the job of my dev server | 22:15 |
| ali1234 | you could just purchase one of those ubuntu cloud boxes ;) | 22:15 |
| shauno | I'm curious why you're aiming for a rack? they only really make sense where you require the density | 22:16 |
| ali1234 | yeah, i'd just go with a tower for this | 22:16 |
| shauno | almost any other scenario a tower is going to be more machine for less money | 22:16 |
| ali1234 | also i wouldn't put computers in a garage, they will rust | 22:18 |
| ali1234 | been there done that | 22:18 |
| ali1234 | same goes for cellars | 22:18 |
| ali1234 | i'd get a tower, put one of these in it: http://www.scan.co.uk/products/icy-dock-mb996sp-6sb-6-sata-hot-swap-525-bay-backplane-raid-cage and a load of these: http://www.scan.co.uk/products/960gb-crucial-m500-25-sata-iii-6gb-s-slim-7mm-ssd-marvell-500mb-s-read-400mb-s-write-80k-iops | 22:20 |
| ali1234 | lots of RAM, and i5 or i7 | 22:21 |
| bigcalm | I either need a quiet yet powerful machine (my laptop is just about up to the job, but I don't want to have it as my dev server). Or a not so quiet machine that I can stick somewhere out of the way | 22:21 |
| ali1234 | that would be quiet if you go i5 ^ | 22:21 |
| ali1234 | and get lots of fans instead of one huge one | 22:22 |
| ali1234 | not silent mind, but you wouldn't hear it from another room | 22:22 |
| bigcalm | Thinking about it, the new house has a utility room that is more part of the house than the garage | 22:22 |
| penguin42 | ali1234: I have a crucial v4 on my desk with bad sectors and less than a year old | 22:24 |
| ali1234 | v4 is like 2 generations old though | 22:25 |
| penguin42 | hmm | 22:27 |
| bigcalm | Hotel booked for the RAT \o/ | 22:28 |
| bigcalm | Looking for bare bones machines, keep being shown ultra compact bricks | 22:53 |
| bigcalm | I'm really out of the game | 22:54 |
| ali1234 | don't buy a barebones | 22:57 |
| ali1234 | why would you even do that? | 22:57 |
| ali1234 | you'll get a crappy loud case with noisy fans and a rubbish PSU if you do that | 22:57 |
| bigcalm | Okay, looking at other things now | 22:59 |
| bigcalm | http://www.ebuyer.com/581296-hp-proliant-ml310e-gen8-724160-035 | 22:59 |
| bigcalm | Though I have no idea what Xeon CPUs are like | 22:59 |
| shauno | if your laptop is ordinarily up to the job, xeon's likely to be overkill. and they rarely hit a nice bang/buck ratio unless you actually need that overkill | 23:00 |
| bigcalm | So they are generally higher performance than the 4th gen i7 I have in this laptop? | 23:01 |
| ali1234 | laptop i7s aren't the same as desktop i7s | 23:02 |
| penguin42 | bigcalm: The single socket xeons are very close indeed to the equivalent spec i7 | 23:02 |
| shauno | now that I think of it, I'm not actually sure what a xeon's selling point is. it used to be dollops of cache, but desktop processors have caught up with that pretty well | 23:03 |
| penguin42 | I thought it was ECC memory | 23:10 |
| penguin42 | although it looks harder to compare these days - no 100% matches to the same model as the xeon | 23:15 |
| popey | evening all | 23:31 |
| popey | uh, morning all | 23:31 |
| penguin42 | hey Popey | 23:31 |
| popey | hows tricks? | 23:31 |
| penguin42 | yeh ok, yourself? | 23:31 |
| bigcalm | Morning popey | 23:31 |
| popey | tickety boo | 23:32 |
| bigcalm | popey: hotel booked | 23:32 |
| popey | which one? | 23:32 |
| bigcalm | Good question | 23:32 |
| popey | holiday inn? | 23:32 |
| bigcalm | Yes, that one | 23:32 |
| bigcalm | £59 quid with Expedia | 23:33 |
| bigcalm | And 8% cash back via quidco | 23:33 |
| popey | oh nice | 23:33 |
| popey | good deal | 23:33 |
| bigcalm | Yeah, lot cheaper than getting a room in Farnham | 23:34 |
| popey | interesting hotel | 23:34 |
| bigcalm | How so? | 23:34 |
| popey | So, Queen Victoria used to get the train from London to Farnborough | 23:34 |
| popey | Then a carriage from the station to that hotel | 23:34 |
| popey | She would stay there | 23:34 |
| bigcalm | I've used Holiday Inn else where in the country. Hopefully the quality is standard | 23:35 |
| bigcalm | Cool | 23:35 |
| popey | And then inspect the troops on the nearby parade grounds | 23:35 |
| popey | it has a lot of history | 23:35 |
| popey | we looked at that place for our wedding reception | 23:35 |
| bigcalm | Big place | 23:37 |
| bigcalm | Quite looking forward to it now | 23:37 |
| popey | yay | 23:38 |
| bigcalm | Next need to book a hotel for oggcamp | 23:39 |
| bigcalm | Which might be the Holiday Inn as well :D | 23:39 |
| bigcalm | Know if anybody has booked hotels yet? | 23:40 |
| bigcalm | I can hear movement up stairs | 23:41 |
| bigcalm | Might be time to go to bed | 23:41 |
| bigcalm | Toodles | 23:41 |
| popey | i have no idea about hotels for oggcamp | 23:44 |
| dutchie | surely the venue itself is the obvious choice | 23:51 |
| popey | yeah, bit spendy though | 23:52 |
| dutchie | i wonder if there are any guest houses in wolvercote | 23:53 |
| dutchie | i guess there are the travelodge and the holiday inn right there | 23:54 |
| * dutchie could probably just stay with postgrad friends in central ox though :D | 23:55 | |
| === Lcawte is now known as Lcawte|Away | ||
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!
