| Time |
Nick |
Message |
| 07:00 |
|
agoben joined #evergreen |
| 07:05 |
|
JBoyer joined #evergreen |
| 07:21 |
|
rjackson_isl joined #evergreen |
| 08:00 |
|
_bott_ joined #evergreen |
| 08:41 |
|
Dyrcona joined #evergreen |
| 08:42 |
* Dyrcona |
searches logs for that link csharp shared about redis. |
| 09:02 |
|
bos20k joined #evergreen |
| 09:14 |
|
collum joined #evergreen |
| 09:21 |
|
rlefaive joined #evergreen |
| 09:25 |
|
mmorgan joined #evergreen |
| 09:28 |
jeff |
csharp, Dyrcona: what's the context? just curious, or looking to solve a perceived problem with memcached? |
| 09:30 |
|
yboston joined #evergreen |
| 09:36 |
jeff |
Dyrcona: if you hadn't found it yet, i think it was this: https://www.infoworld.com/article/3063161/nosql/why-redis-beats-memcached-for-caching.html -- promotional piece written by a Redis Labs employee and published in InfoWorld's "New Tech Forum", for which ``InfoWorld does not accept marketing collateral for publication and reserves the right to edit all contributed content.'' |
| 09:44 |
|
jvwoolf joined #evergreen |
| 09:46 |
Dyrcona |
jeff: Yeah, that's the one I read. |
| 09:46 |
|
kmlussier joined #evergreen |
| 09:47 |
miker |
jeff: heh... thanks for the context, there ;) |
| 09:47 |
Dyrcona |
I was just curious, and even after reading that, I'm not sure how much (if at all) Evergreen would benefit by switching to redis. |
| 09:48 |
Dyrcona |
I might give redis a spin some day to see if all those features (and complication) are worth it. |
| 10:02 |
|
mmorgan1 joined #evergreen |
| 10:09 |
miker |
there's not a ton of benefit to switching for how we use memcache today. redis would allow some more sophisticated uses, such as coalescing sets (redis hashes) and then updating just parts -- and and expiring the whole thing. the locale-specialized org trees come to mind, and the per search facet cache. but it's not pressing |
| 10:26 |
Dyrcona |
miker: I didn't think in specifically those terms, but I figured there would be some minor benefit, but nothing gigantic. |
| 10:35 |
|
mmorgan joined #evergreen |
| 11:10 |
|
mmorgan left #evergreen |
| 11:31 |
|
Christineb joined #evergreen |
| 11:33 |
|
kmlussier joined #evergreen |
| 12:26 |
|
khuckins joined #evergreen |
| 12:41 |
|
jihpringle joined #evergreen |
| 12:53 |
Dyrcona |
@dunno |
| 12:53 |
pinesol_green |
Dyrcona: Down time is a fact of business when you're a poor 501c3 corporation. |
| 13:18 |
jeffdavis |
bug 1729922 was supposed to be fixed in 3.0.3 and is marked as Fix Released, but it looks like commit 74dc248 never made it into a release - it's in master, but the fix is not in tags/rel_3_0_3 or tags/rel_3_0_4 |
| 13:18 |
pinesol_green |
jeffdavis: [evergreen|Galen Charlton] LP#1729922: correctly display most recent transit in item status - <http://git.evergreen-ils.org/?p=Evergreen.git;a=commit;h=74dc248> |
| 13:18 |
pinesol_green |
Launchpad bug 1729922 in Evergreen "Web Client: Most Recent Transit is showing older transit" [Medium,Fix released] https://launchpad.net/bugs/1729922 |
| 13:20 |
jeffdavis |
gmcharlt: ^ |
| 13:22 |
gmcharlt |
jeffdavis: hmm, I confirm that it didn't make it into rel_3_0 |
| 13:24 |
gmcharlt |
cherry-picks cleanly, so I've now added it to rel_3_0 and changed the target to 3.0.5 |
| 13:24 |
jeff |
jeffdavis++ gmcharlt++ |
| 13:24 |
jeffdavis |
gmcharlt++ |
| 13:44 |
Dyrcona |
jeffdavis++ gmcharlt++ |
| 13:44 |
Dyrcona |
django-- |
| 13:45 |
jeff |
Dyrcona: Syrup troubles? |
| 13:45 |
Dyrcona |
So, I've apparently arrived at a point where I need to learn some Django and change code. |
| 13:45 |
Dyrcona |
jeff: Yes. |
| 13:46 |
csharp |
jeff: basically curious - particularly interested in the high-availability/replication features of redis |
| 13:46 |
Dyrcona |
I've patching each error as they come up, but this one looks like it will require new code: ImportError: No module named transaction |
| 13:46 |
csharp |
currently have two memcache servers and would like to be able to "roll one out" without the world crashing down |
| 13:47 |
jeff |
csharp: because Evergreen sessions aren't currently durable and loss/restart of a memcached instance is therefore painful? :-) |
| 13:47 |
Dyrcona |
Well, for one. :) |
| 13:48 |
csharp |
jeff: zactly |
| 13:48 |
jeff |
Dyrcona: what version of Django are you targeting, and for any particular reason? |
| 13:48 |
* Dyrcona |
notes that upgrading a program from Qt3 to Qt 5 was not this painful. |
| 13:48 |
Dyrcona |
Actually, it stated with Qt 2. |
| 13:49 |
Dyrcona |
Django 1.8 because it comes in the package. |
| 13:49 |
berick |
csharp: i too feel the pain of not being able to detach a memcache server. related, i suspect there are tools to make that a possibility, but i haven't looked. |
| 13:51 |
jeff |
my preference for that is to treat memcached like a cache, and have sessions backed by something more durable, like postgres (either in the main db or a different db) or even... redis. :-) |
| 13:51 |
* jeff |
looks to see if there is a bug on that wish yet |
| 13:56 |
|
kmlussier joined #evergreen |
| 14:00 |
|
alynn26 joined #evergreen |
| 14:09 |
miker |
jeff: now that we require a version of pg that has unlogged tables, we could do that transparently, I believe |
| 14:10 |
* berick |
reads about unlogged tables |
| 14:10 |
jeff |
bug 1753565 to start |
| 14:10 |
pinesol_green |
Launchpad bug 1753565 in Evergreen "Support durable authentication sessions" [Wishlist,New] https://launchpad.net/bugs/1753565 |
| 14:15 |
* Dyrcona |
has used unlogged tables for some db updates. |
| 14:15 |
csharp |
jeff++ |
| 14:16 |
|
jvwoolf joined #evergreen |
| 14:22 |
JBoyer |
Hey Acquisitions users, does there seem to be an upper limit to how many items you can have on a PO before it will start to fail to Activate? We had a 115 item order that refused to activate multiple times. When it finally did it took several minutes. |
| 14:23 |
jeff |
csharp: is your HA/replication interest mostly around auth sessions, or other things? |
| 14:24 |
jeff |
csharp: i suppose with the current situation in which memcached change/restart invokes an authpocalypse, the other things that might not be durable/cached-only may not actually be noticed. :-) |
| 14:25 |
JBoyer |
jeff, re: bug 1753565, there's at least one site running multiple servers. :) I've not changed the server list midday though so I've not been bothered by the lack of replication. |
| 14:25 |
pinesol_green |
Launchpad bug 1753565 in Evergreen "Support durable authentication sessions" [Wishlist,New] https://launchpad.net/bugs/1753565 |
| 14:25 |
csharp |
jeff: auth sessions, yes |
| 14:26 |
jeff |
JBoyer: data point! thanks! |
| 14:27 |
Dyrcona |
Hm. No Launchpad for Syrup. Not that we need one. No one has touched it in 3.5 years. |
| 14:28 |
Dyrcona |
What JBoyer said goes for us, too. |
| 14:29 |
Dyrcona |
And, I'm planning to set up new ones and take the old ones out of service. |
| 14:29 |
Dyrcona |
memcached servers that is. |
| 14:30 |
jeff |
JBoyer: I suspected that we had at least one site with single memcached clients hitting multiple servers (and therefore hashing to distribute keys), but don't remember trying it myself. In theory, even if we have client implementations that don't share a common hashing algorithm, oils_auth_* keys are only touched by one implementation, etc... |
| 14:30 |
Dyrcona |
No working repo for Syrup, either, but MassLNC has a repo. I'll make sure that I have access to that. |
| 14:30 |
jeff |
Dyrcona: I shared our changes with you a while back, right? |
| 14:30 |
* jeff |
looks |
| 14:30 |
jeff |
oh, looks like just the one superpage change. oops! |
| 14:31 |
Dyrcona |
Syrup uses so much that has been removed since Django 1.4.... |
| 14:34 |
miker |
re multi-memcache EG instances, ditto here, plenty of them. (IMO, if you're restarting memcache during normal operation there are some service layout issues to address. hope that doesn't sound /too/ judge-y ;) ) |
| 14:36 |
miker |
(jeff: and we (EG folken) went out of our way to make sure that at least the C and perl bindings were using identical hashing algos, for that reason) |
| 14:38 |
jeff |
yeah, we are not in the habit of restarting memcached during normal operation. :-) |
| 14:40 |
berick |
well, the problem is normal operation is 24/7 |
| 14:42 |
jeff |
and good to hear that the hashing implementations are in sync. i could have gone looking / checked notes from the last time i dug into that. :-) |
| 14:42 |
berick |
hm, i thought they weren't :( |
| 14:43 |
berick |
been a while since I looked too |
| 14:43 |
* jeff |
laughs |
| 14:43 |
berick |
i know at one point i tried pull auth tokens from Perl and they were missing, but that was several years back |
| 14:45 |
berick |
i wouldn't be surprised if the package maintainers coalesced around a sane default over the years. |
| 14:46 |
jeff |
also, in past experience early eviction tended to eat persist tokens before their TTL. i think this may vary depending on how your memcached has allocated things. |
| 14:46 |
gmcharlt |
dbwells: speaking generally, do you agree that bug 1699838 is reasonably targetted for 3.1-rc? |
| 14:46 |
pinesol_green |
Launchpad bug 1699838 in Evergreen "OU setting needed for default pickup location for staff placed holds" [Medium,Confirmed] https://launchpad.net/bugs/1699838 |
| 14:47 |
jeff |
but then i'm talking from memory without verifying again. |
| 14:48 |
jeff |
also, memcached has an external storage shim as of late last year, but it's designed for things where your values are large, not general persistence of small values. |
| 14:51 |
Dyrcona |
And the syrup/masslnc.git repo has its first branch. |
| 14:51 |
Dyrcona |
Not counting master, of course. |
| 14:51 |
Dyrcona |
Still a work in progres. |
| 14:53 |
kmlussier |
Dyrcona++ |
| 14:55 |
kmlussier |
JBoyer: I'm having template trouble on a system where I previously had webstaff copy templates. I remove the webstaff template setting from actor.usr_setting. Is that all that needs to be done to get the xul templates on my user account to convert the next time I log into the web client? |
| 14:56 |
miker |
berick: hrm... I thought we /did/ align the hashing. IIRC, we had to switch out the C lib we were using. but, in reality, it doesn't matter because we don't read auth sessions from perl, and we don't read anything else from C ;) |
| 14:57 |
dbwells |
gmcharlt: I think that is fine, yes. I have targeted it to the RC. |
| 14:57 |
JBoyer |
kmlussier, you'll also want to either logout and close the browser, OR just remove the cat.copy.templates.usr key from Session storage before reloading the vol/cat editor. |
| 14:57 |
gmcharlt |
dbwells: thanks! |
| 14:58 |
kmlussier |
JBoyer: OK, I think I logged out, but I'll try again. |
| 14:59 |
kmlussier |
JBoyer: Actually, now that I think about it, I tried a different browser on my 2nd or 3rd attempt, so I think that would have covered removing it from Session storage. |
| 15:00 |
JBoyer |
Ah, yeah. That should have covered that. |
| 15:00 |
JBoyer |
What kind of things are you seeing? |
| 15:02 |
kmlussier |
I'm getting an empty dropdown when I click to view templates. There is one xul client template in my usr settings that I'm hoping to see. |
| 15:03 |
kmlussier |
JBoyer: Sigh, now I'm seeing it. |
| 15:04 |
JBoyer |
There is also the fact that it seems to sometimes require a reload before actually appearing. :/ |
| 15:05 |
kmlussier |
When I finally got it to work, I was accessing templates from the Local Admin / Volume/Copy Template Editor instead of going through the Volume/Copy Editor. Would that make a difference? |
| 15:10 |
dbs |
Dyrcona: I think you can just treat https://git.evergreen-ils.org/?p=Syrup.git as a working repository and extend permissions to that |
| 15:11 |
dbs |
I don't think there needs to be "core" vs "working" repos |
| 15:11 |
dbs |
(not for Syrup anyway) |
| 15:13 |
Dyrcona |
OK. I'm using the masslnc repo for now. |
| 15:21 |
JBoyer |
kmlussier, it shouldn't but I can't say for certain that it doesn't. :/ |
| 15:23 |
|
rlefaive joined #evergreen |
| 15:35 |
kmlussier |
JBoyer: Yeah, I tried reproducing my steps again, and I don't think the interface matters. I'm not quite sure what the magic sauce is yet. |
| 15:41 |
kmlussier |
Ok, on this last attempt, I finally got it to load after clearing the browser cache and reloading. But I had previously cleared the browser cache and reloaded without success. :/ |
| 15:41 |
kmlussier |
@eightball Does Evergreen hate me? |
| 15:41 |
pinesol_green |
kmlussier: The outlook is hazy, please ask again later. |
| 15:45 |
Dyrcona |
@eightball Does Django hate me? |
| 15:45 |
pinesol_green |
Dyrcona: The answer is def-- oooh! shiny thing! |
| 15:50 |
kmlussier |
@eightball Do you ever provide clear answers? |
| 15:50 |
pinesol_green |
kmlussier: The outlook is poor. |
| 15:50 |
bshum |
Haha |
| 15:50 |
kmlussier |
OK clearing the browser cached seems to be consistently working for me now. I'm going with it. |
| 15:51 |
kmlussier |
cache even |
| 15:58 |
Dyrcona |
wsgi-- |
| 16:00 |
Dyrcona |
Nope. wsgi++ |
| 16:01 |
Dyrcona |
django-- # Still some Django foolishness. |
| 16:01 |
Dyrcona |
pinesol_green: Django does hate me. |
| 16:01 |
pinesol_green |
Dyrcona: did you finish your beer? |
| 16:01 |
Dyrcona |
No, but I'll have a glass of red wine, nothing too expensive, a nice blend for California will do. |
| 16:02 |
|
frank_mx joined #evergreen |
| 16:03 |
frank_mx |
Hi all, Has someone used "Authentication Proxy" with gmail apps for work? Is there any kind of documentation about it? |
| 16:04 |
alynn26 |
Has anyone had any luck getting Dymo Labelwriters to work with the Webclient? |
| 16:06 |
Dyrcona |
And, I've finally gotten past 500 server errors into a Django import error. |
| 16:07 |
kmlussier |
progress++ |
| 16:07 |
alynn26 |
been getting the same error as bug 1741299 |
| 16:07 |
pinesol_green |
Launchpad bug 1741299 in Evergreen "DYMO LabelWriter 450 Turbo Hatch Label print fails with margin error" [Undecided,New] https://launchpad.net/bugs/1741299 |
| 16:07 |
Dyrcona |
I don't know if I'd really call this progress. :) |
| 16:12 |
|
khuckins joined #evergreen |
| 16:13 |
Dyrcona |
Oh, nice. Syrup overrides a method removed in Django 1.7. |
| 16:15 |
Dyrcona |
But, that is tomorrow's starting point. |
| 16:47 |
|
jvwoolf left #evergreen |
| 17:01 |
|
rlefaive joined #evergreen |
| 19:29 |
pastebot |
"Bmagic" at 64.57.241.14 pasted "Why did SIP not send back the 24 message?" (75 lines) at http://paste.evergreen-ils.org/531 |
| 20:55 |
jeff |
presumably because "request returned no data : open-ils.cstore.direct.money.open_user_summary.retrieve" |
| 22:25 |
Bmagic |
jeff: I finally found it. The /openils/var/log/std-err.log helped |
| 22:25 |
Bmagic |
it was complaining about Sip/Configuration/Institution.pm line 45 not being a hash |
| 22:25 |
Bmagic |
I found a flaw in the oils_sip.xml config related to relais_extensions_to_msg24 |
| 22:26 |
jeff |
oh? |
| 23:23 |
Bmagic |
jeff: it was as flaw with my settings. Not EG. |
