| | <<<Back 1 day (to 2014/11/30) | 20141201 |
| kens | I see Helmut is wondering why a file wit a font not embedded render differently on different platforms again. | 14:16.20 |
| | Oh and he hasn't sent teh 2 fonts, excellent..... | 14:17.49 |
| chrisl | I think I have made changes that could affect that sort of thing, but I'm not in a hurry to look at it | 14:21.21 |
| kens | Well it works just fine when substituting Times-Roman. I don't jhave a copy of andalus on my system, and even if I did, we'd still need to see the ones he's actually using. Plus, of course, he doesn't seem to grasp that when we substitute for a missing font, all bets are off...... | 14:22.21 |
| | Actually, I take it back, I *d* have a copy of andalus | 14:23.09 |
| | chrisl looks like exactly the same problem, but a different glyph. And he still isn't sending the fonts. | 14:52.04 |
| chrisl | kens: same problem as what? | 14:52.24 |
| kens | THe apostrophe seems to be the same problem as the Euro. Substituting a TTF on disk for missing fonr in PDF, different results on different systems | 14:53.02 |
| chrisl | Oh, right. It's strange he *claims* 9.10 works..... | 14:53.31 |
| kens | He keeps on claiming 9.10 works, for pretty much everything | 14:53.46 |
| | But without the fonts he's using, how cna we tell ? | 14:53.55 |
| chrisl | Well, so far, the only case I've actually *seen* work in 9.10 and not in 9.15 was the GS_LIB issue | 14:55.12 |
| kens | Indeed. | 14:55.24 |
| henrys | there were a few more updates to the agenda let me know if you want me to send the diffs. | 15:06.09 |
| jogux | marcosw: I think we're fighting for control of the win8 box? :) | 18:04.12 |
| | maybe I mean 'hope'... | 18:04.20 |
| marcosw | jogux: are you fiddling with the win81-ats client? It's not running and when I try to start the client I get a "*** Failed to init the client" message. I'd like to reboot the VM. | 18:05.25 |
| jogux | yeah. go for a reboot | 18:05.46 |
| | the problem is the ramdisk vanished when I tried to reboot it | 18:05.52 |
| | and a reboot didn't fix it | 18:05.58 |
| | (this all started because I was trying to fix a bug Robin ran into) | 18:06.15 |
| marcosw | weird I haven't changed anything that could cause that (I made some changes to the cpu allocation for one the artifex cluster VM, but that's not part of the ATS). | 18:07.05 |
| jogux | yeah. it only started when I rebooted it to check my new version of the client was correctly installed etc :-( | 18:07.44 |
| marcosw | btw, I did add code to autostart the ATS client, so now a coldboot of toast and scones will (should) bring everything back up. | 18:08.14 |
| jogux | marcosw: yeah, I noticed that - hence trying to check I hadn't broken it :-) | 18:08.29 |
| | Hmmm. Did I buy a license for the ramdisk? how long was the trial...? | 18:08.43 |
| marcosw | I should probably step back and let you figure out what's going on, my knowledge of win8 is not even enough to be dangerous. | 18:09.51 |
| jogux | ok. I think it may well be that the ram disk trial expired. oops. let me login and confirm that. | 18:10.35 |
| marcosw | that's an easy fix, just throw money at it :-) | 18:10.53 |
| jogux | yes. the trial has expired. I'm stupid. | 18:12.16 |
| | marcos: it's a USD thing, so it may be easiest for your to deal with this? http://winramtech.hostei.com/RAMDriv/ (saves a couple of USD -> GBP -> USD -> GBP cycles :) ) - I think we want the $12 one. | 18:13.32 |
| marcosw | jogux: sure. | 18:13.43 |
| | we don't want the $25 one that includes updates? | 18:14.51 |
| jogux | marcosw: other links: http://twiki.ghostscript.com/do/view/GhostDocs/ATSClientWindows8 and http://winramtech.hostei.com/RAMDriv/main/ramdiskfull.htm has the "how to update" bit that presumably we'll need to follow once they send us the non-trial driver. | 18:15.12 |
| | marcosw: I'll leave the decision to you :) I have no objection. | 18:16.00 |
| marcosw | I think we can afford the $25 (we've spent nearly that much in engineering resources that just discussing it). | 18:17.01 |
| jogux | :-) | 18:18.37 |
| | it is very cheap | 18:19.02 |
| marcosw | I've received the paypal receipt but not the software. If it doesn't arrive in 48 hours I'm supposed to send them a message. | 18:20.33 |
| jogux | okay. actually I'll install the latest trial, that'll get it going again for Robin... | 18:23.00 |
| Robin_Watts | thanks guys, | 18:23.18 |
| jogux | the load average on toast just hit 21. it's not been overly responsive just now :) | 18:36.32 |
| marcosw | jogux: the non-expiring ramdrive arrive, I'll forward you the email. | 18:38.03 |
| jogux | oh, great. I'm still struggling to update the trial. I think windows won't update the driver because the version hasn't changed... so need to uninstall/resinstall. | 18:40.12 |
| marcosw | I wonder how the load average got to 21. I didn't think the CPU resources are that overcommitted. | 18:40.19 |
| | that's why I sent you the non-expiring one, figured no point in going through the exercise twice :-) | 18:40.42 |
| jogux | :) | 18:41.17 |
| | there was an rsync running, so I'm guessing it may have been i/o bound. | 18:41.26 |
| | hm, no email yet :-S | 18:43.21 |
| marcosw | I've stopped the rsync, was doing a pre-backup of the vm images (figured I'd do that and then re-run the rsync after I take them down, but I should wait until tonight). | 18:44.23 |
| | I'll put the ramdisk onto the win8.1 box's desktop. | 18:45.26 |
| jogux | oh, I meant to reply to your email: there is a way to cleanly shutdown ATS; iirc something like if you SIGHUP scheduler.pl it'll stop running new jobs... give ats 20 minutes or so for existing jobs to finish then you can shutdown VMs and are guaranteed not to cause any jobs to fail. | 18:45.29 |
| | that would be great, thanks | 18:45.35 |
| marcosw | silly to email you the file only to have you ftp it back. | 18:45.39 |
| jogux | :-) | 18:45.42 |
| marcosw | actually I'm not sure how to get stuff onto the win8 box, so I've put it on toast in root's home. | 18:47.11 |
| | jogux: great, I'll add the shutdown instructions to the twiki (they might be there already, but I'll add them to where I can find them). | 18:47.47 |
| jogux | okay | 18:48.53 |
| | can you let me know when the ramdisk is there? | 18:48.58 |
| marcosw | it should be on toast now. | 18:49.07 |
| jogux | where on toast? :-) | 18:49.45 |
| marcosw | ~root | 18:50.00 |
| jogux | got it, ta. | 18:50.37 |
| | right, that worked. yay. | 19:01.40 |
| Robin_Watts | so I can rerun the windows jobs? | 19:02.51 |
| jogux | yeah, go for it | 19:03.23 |
| | I already re-ran one of yours to try and see if my changes are actually working/deployed right. | 19:03.35 |
| | and it appears they are. woot. | 19:03.51 |
| Robin_Watts | Thanks. | 19:06.05 |
| rayjj | marcosw: I'm having trouble reproducing the segfaults that I got on my most recent "ray" cluster test. I looked at the logs and copied/pasted the command line and I can't get it to fail on peeved (release or debug build) | 20:01.52 |
| | marcosw: I've tried a half dozen or so, and they all complete. | 20:02.27 |
| | marcosw: do you have any idea what I might be doing wrong, or how I can reproduce the segfault (signal 11) | 20:04.25 |
| | I was hoping to wrap up these tmpfile + BGPrint bugs before leaving for the meeting | 20:04.56 |
| | maybe I need to just automate trying them from the email list (has anyone done that?) | 20:06.19 |
| marcosw | marcosw: I'll take a look after I finishing entering a bug. | 20:09.34 |
| rayjj | marcosw: If that was to me, thanks. No rush, but you have more experience in the cluster | 20:14.24 |
| marcosw | rayjj: yeah, it was to you, not sure what I was thinking... | 20:14.46 |
| rayjj | just talking to yourself online :-) | 20:15.07 |
| marcosw | typing to oneself, wonder if that will replace talking to oneself. | 20:15.50 |
| rayjj | marcosw: I know people that text to themselves to say things like "pick up milk on the way home" | 20:16.40 |
| Robin_Watts | "remember to check for text messages" | 20:17.02 |
| rayjj | talking to oneself is too immediate and ephemeral -- no log unless you have a surveillance app running all the time | 20:17.28 |
| | Robin_Watts: thanks for reminding me. I've often found text messages > 12 hrs old laying around unread :-/ | 20:20.23 |
| marcosw | rayjj: using the binary built from your latest clusterpush I see a segfault on peeves with the following command line: | 20:33.08 |
| | ./gs/bin/gs -dBGPrint=true -o test.pbm -dMaxBitmap=10000 -sDEVICE=pbmraw -r300 ~/cluster/tests/pdf/Bug6901014_CityMap-evince.pdf | 20:33.08 |
| | I'm running this from the directory: /home/marcos/cluster/users/ray/ghostpdl | 20:33.31 |
| | This is what the autogen.sh command looks like: | 20:34.10 |
| | ./autogen.sh "CC=gcc -m64" --disable-fontconfig --without-system-libtiff --without-libpaper --disable-dbus --without-libidn | 20:34.12 |
| | rayjj: the failure is not related to the autogen.sh options. The copy of your latest clusterpush source in ~marcos/ghostpdl.ray was build with "./autogen.sh ; make" and also fails with the given command line. | 20:46.42 |
| rayjj | marcosw: thanks. I'm going to try that in my directory with my build... | 20:59.49 |
| | marcosw: I don't see ~marcos/ghostpdl.ray | 21:02.49 |
| marcosw | pwd | 21:03.27 |
| | on peeves? | 21:03.47 |
| | marcos@peeves:[92]% pwd | 21:03.51 |
| | /home/marcos/ghostpdl.ray | 21:03.51 |
| rayjj | marcosw: I'm on peeved. Just a sec | 21:04.09 |
| | hmm... it's there on peeves. | 21:04.36 |
| | trying the command now | 21:04.40 |
| marcosw | I'm off for a bit to run errands and lunch. | 21:05.08 |
| rayjj | marcosw: OK. I get segv there. Thanks | 21:05.57 |
| | decent news (I guess) I can get the segfault with a debug build | 21:12.59 |
| | next to try it on my linux VM ... | 21:14.44 |
| | Thanks again, marcosw (for the logs) | 21:14.57 |
| | I don't know why it wasn't failing for me, but without having my old command line, ??? | 21:15.42 |
| | that same command line is fine on my windoze build | 21:16.01 |
| marcosw | I'm going to take down the Fremont ATS in a bit to back it up and perform some other maintenance. It will be down for a few hours but I don't believe anyone in a timezone where it's still business hours will need to use it. | 22:57.01 |
| | Forward 1 day (to 2014/12/02)>>> | |
IRC Logs