IRC Logs for #crux-devel Thursday, 2015-10-15

Workster# adding harfbuzz for improved OpenType features auto-hinting00:28
Workster# introduces a cycle dep to harfbuzz depending on freetype wanted by upstream00:28
Worksteri've hit that many times on chroots and docker containers00:28
Worksterreally annoying really.00:28
jaegerDoesn't seem like there's much choice. Disable it or live with it, I suppose00:41
Worksteri dunno do we want freetype improved auto-hinting? could just build the iso without harfbuzz but leave it as a optional dep in ports.00:51
Worksterwonder what upstream has to say between freetype and harfbuzz and what was the other port that needs harfbuzz pango ?01:01
Worksterbbl lunch01:01
*** ________mavric has quit IRC02:48
*** ________mavric has joined #crux-devel02:49
frinnstjaeger: iso/master - iso/grub.cfg07:48
frinnstmenuentry "Install/Upgrade CRUX 3.1 (UEFI mode)" {07:48
Romsterthen do a sed in Makefile08:18
*** jue has quit IRC08:33
*** jue has joined #crux-devel08:42
*** Romster has quit IRC10:13
*** Romster has joined #crux-devel10:18
*** nomius has quit IRC10:57
*** nomius has joined #crux-devel10:57
jaegerfrinnst: already fixed that one locally, the rc should have the right version number13:42
jaegerI can be pretty slow about pushing my local changes, I know... I just figure most of the time nobody's using the iso git tree besides me13:49
jaegerand it's a bit of a moving target around release time13:49
frinnstjaeger: 16:19 <+nasbot> baljar: "dedup" - DEDUP IS HAZARDOUS. Where ZFS recommends 1Gb ram per 1Tb of HDD space, dedup REQUIRES 32Gb ram per 1Tb to not choke on larger tasks (like importing a pool that has suffered power failure, or destroying a populated dataset). The first rule of dedup, is don't dedup. The second rule of dedup, is don't dedup. The third rule of dedup, is don't use snapshots.14:32
frinnstthought of you :D14:32
jaegeramen to that, heh14:33
jaegerI wish that had been public/documented knowledge back when this was my problem :P14:34
jaegerwe would need like 6TB of RAM14:34
frinnstbtrfs seems a lot more resource friendly than zfs14:35
jaegernormal ops are fine, resource-wise... dedup+snapshots is the devil14:36
frinnstwe save 41% with deduplication on our production systems14:40
frinnstmost servers are just windows 2008r2 or 2012r214:42
rmullI'm currently having strange IO waits when doing simply cp/mv operations on a btrfs RAID117:37
rmullHangs for like 10s before completing17:37
*** heroux has quit IRC17:58
*** heroux has joined #crux-devel17:59
frinnstbut try this: alias cp='cp --reflink=auto'18:07
frinnstI remember similar issues waaaay back. but that was years ago18:08
rmullWell, I checked dmesg and it's full of read errors. So that's probably the issue. UGH19:46
*** Lukc has joined #crux-devel20:18
*** jue has quit IRC20:29
*** jue has joined #crux-devel21:57
frinnstwhat kind of errors?22:07
rmullfrinnst: Here's a sample - I guess I need to replace sdc.
rmullI'll probably try re-plugging the sata cable and stuff first though22:25
*** dlcusa has joined #crux-devel23:21
dlcusaI noted a 3.2-rc1 UEFI prob on #crux--was that the right place?23:22
frinnstyeah doesnt look like a btrfs issue23:24
frinnstbut isnt these awesome? BTRFS: read error corrected: ino 19171208 off 786432 (dev /dev/sdc sector 418760880)23:25
rmullYeah, I am not too angry about seeing those, haha.23:26
*** dlcusa has left #crux-devel ("Leaving")23:30

Generated by 2.11.0 by Marius Gedminas - find it at!