Time |
Nick |
Message |
06:00 |
pinesol |
News from qatests: Testing Success <http://testing.evergreen-ils.org/~live> |
07:09 |
|
rjackson_isl joined #evergreen |
08:05 |
|
Dyrcona joined #evergreen |
08:22 |
|
rfrasur joined #evergreen |
08:36 |
|
mmorgan joined #evergreen |
08:38 |
|
mantis joined #evergreen |
09:09 |
|
alynn26_away joined #evergreen |
09:12 |
stephengwills |
Note to self: volunteer more with DIG For others doing upgrades to their database, “psql -a --echo-all -f ~/upgrade_logs/version1-version2.sql. &> upgradelog. made a huge difference for me. Also, the v6 and above documentation says: (Debian and Ubuntu) No extra perl modules required for these distributions. this isn’t correct for 3.7 which requires String::KeyboardDistance and Text::Levenshtein::Damerau::XS module. Balsa |
09:12 |
stephengwills |
now running 3.7.1 thanks for the support yesterday. |
09:12 |
|
phasefx_ joined #evergreen |
09:13 |
alynn26_away |
stephengwills++ DIG++ |
09:18 |
Dyrcona |
stephengwills: I make it a rule of thumb to always run the prerequisite install steps when doing "major" version upgrades, i.e. from 3.X to 3.Y. Our first version digit is pretty much meaningless. |
09:21 |
Dyrcona |
stephengwills: Also, which upgrade documentation are you talking about? The server upgrade document in the git repo? |
09:23 |
stephengwills |
no, at docs.evergreen-ils.org. and I thought I did make -f Open-ILS/src/extras/Makefile.install postgres-server-ubuntu-bionic-10 on my postgresql server. isn’t that enough to install the prerequisites? |
09:24 |
Dyrcona |
stephengwills: Maybe not. It could be a bug in the prerequisite install target. |
09:25 |
stephengwills |
i’ll try and cut some time to poke around more before I get complacent with my installed server and move on to other stuff. again, so helpful y’all. thanks. |
09:28 |
Dyrcona |
To be honest, I almost never refer to the documentation and go by release notes and trial and error. I've usually done the upgrade several times on test VMs before I do the production upgrade. |
09:30 |
stephengwills |
in the past I’ve done that too but I am an applications layer guy who spends most of my time in PHP so when I crawling onto the cloud the more NASA-like my steps are the smoother it goes for me. LOL! |
09:31 |
stephengwills |
Note to self: check to make sure the doc.eg-ils.org clearly states i must plug in my computer before I flip the ON switch. |
09:32 |
Dyrcona |
So, you could argue that it is a bug in the Pg server targets. They don't include any CPAN modules, and the Levenshtein, et al., are installed via CPAN. |
09:33 |
Dyrcona |
stephengwills: I arrive at the NASA like steps via trial and error. My upgrades are "scripted." |
09:33 |
stephengwills |
** thumbs up for scripts ** |
09:35 |
Dyrcona |
Sometimes the script are copy and paste from a document. Sometimes the scripts are actual bash scripts. Most of the time, it's a combination with bash for "subsystems." |
09:38 |
Dyrcona |
Here's an example from our most recent upgrade from 3.2 to 3.5: https://docs.google.com/document/d/16DUI57A7zDkfUYwpZC0CLGo_PZO8-K6KGnfWKhwkO-o/edit?usp=sharing |
09:39 |
stephengwills |
nice, thanks |
09:41 |
Dyrcona |
I used more scripts in that one. The document usually has a lot more commands to copy and paste. |
09:43 |
stephengwills |
the insight into your process is helpful. |
09:49 |
Dyrcona |
stephengwills: Lp 1936662 if you'd like to confirm |
09:49 |
pinesol |
Launchpad bug 1936662 in Evergreen 3.7 "Missing Did You Mean Prequisites in Pg Server Installation Make Targets" [Undecided,New] https://launchpad.net/bugs/1936662 |
09:53 |
Dyrcona |
Hm... After looking more closely, they are there. I should double check if they're getting installed or not. It could be that the CPAN_MODULES_PGSQL is just not being called. In which case MARC::File::XML would also be missing on a new installation. |
09:55 |
Dyrcona |
Y'know what. I was too quick to open the Lp bug. They *should* be getting installed. |
09:55 |
Dyrcona |
I'm going to check on a clean VM before I mark the bug invalid, though. |
09:57 |
stephengwills |
Dyrcona++ |
10:14 |
|
jvwoolf joined #evergreen |
10:35 |
Dyrcona |
stephengwills: Did you mention what O/S you were using Ubuntu Bionic (18)? |
10:36 |
stephengwills |
I implied it when quoting the make install for ubuntu bionic and Pg10 |
10:36 |
Dyrcona |
Yeahp. |
10:37 |
Dyrcona |
I have confirmed that the CPAN modules don't get installed on Debian Buster. |
10:37 |
Dyrcona |
I should test again with a Bionic VM. |
10:37 |
Dyrcona |
I may have to amend my bug report. |
10:39 |
Dyrcona |
Maybe focal and stretch, too. Guess I'm spending the weekend rebuilding my local test VMs..... |
10:40 |
stephengwills |
my slice is an AWS EB2 Ubuntu 18.04 LTS with Pg v10 and I did make the extras from Eg3.7.1 on it and, ultimately had to install the two modules I mentioned above to make 3.6.2-3.7.0-upgrade-db.sql happy. |
10:45 |
Dyrcona |
stephengwills: Is the database on a separate host/instance from the Evergreen servers? If it was all on the same machine and the general ubunbut-bionic prerequisites were installed, then the modules should have been there. |
10:50 |
jeff |
I do really wish that Chrome had an option to allowlist an extension (Hatch) without force installing it. |
10:51 |
jeff |
I want Hatch installed where it's been installed, and I want Chrome to not say "oh hey, this weird extension has been installed on this computer, do you want to disable it?" |
10:51 |
jeff |
so far, I haven't seen a way to do that. |
10:55 |
stephengwills |
Drycona my db is on separate postgresql instance |
10:59 |
stephengwills |
I think remember years ago we discussed documenting the postgresql server build as being out of scope for Evergreen Docs. It’s still useful to me to have some of these tricks documented. i.e. the —echo-all switch etc. |
11:00 |
Dyrcona |
stephengwills: That's what I thought. I'm going to leave the bug open and check the currently supported distros to see which ones fail to install the CPAN modules. |
11:01 |
stephengwills |
<3 |
11:09 |
|
jihpringle joined #evergreen |
11:48 |
|
mmorgan1 joined #evergreen |
11:50 |
JBoyer |
jeff, I'm pretty sure that's intentionally impossible for security reasons. |
11:51 |
JBoyer |
Though I suspect if Google wanted to do so they already have a way. |
11:56 |
* jeff |
raises a skeptical eyebrow |
11:57 |
jeff |
what kind of security reasons do you have in mind? |
12:02 |
jeff |
looking at it from another direction, if I can silently force install an extension, why can't I exempt a specific extension from certain warnings if/when installed? |
12:08 |
JBoyer |
Unless I've misunderstood something this is what I'm envisioning as the security explanation: you install rando app xyz, it silently installs a chrome plugin just like we do, but this plugin sends all of your browsing history or whatever to a third party. |
12:09 |
Dyrcona |
jeff: Because. |
12:09 |
JBoyer |
Now, I had initially missed the part about an explicit allowlist, which you're right, would make a lot of sense from a sysadmin pov. |
12:10 |
JBoyer |
(I took your question more like "I intentionally installed this thing that came with a chrome extension, why is chrome making sure I actually want it installed?" |
12:12 |
JBoyer |
I do think it's odd that end users are allowed to disable sysadmin force-installed extensions, but the part about notifying them is likely has a privacy / security rationale. |
12:23 |
jeff |
I don't think that end users can disable sysadmin force-installed extensions. I'm also fairly certain that they aren't notified of them in the way that I'm wanting to disable. If I wanted to force install the extension, I'm pretty sure I'd be set. |
12:24 |
jeff |
But yeah, the idea is not that any old extension has been installed, but that an extension that an administrator has explicitly allowed (but not forced) by extension id has been installed. |
12:24 |
jeff |
Might give things another try and see if the logic has changed, or might open a bug. |
12:25 |
jeff |
(Chrome bug, not Hatch bug) |
12:37 |
|
stephengwills joined #evergreen |
13:19 |
|
mantis joined #evergreen |
13:27 |
|
mantis joined #evergreen |
13:30 |
|
Keith_isl joined #evergreen |
13:31 |
|
mantis joined #evergreen |
13:56 |
JBoyer |
Ah, ok. I think I made some faulty assumptions about what you were doing. |
14:24 |
|
jihpringle81 joined #evergreen |
14:31 |
|
mantis joined #evergreen |
14:35 |
|
mmorgan1 joined #evergreen |
14:42 |
|
rjackson_isl_ joined #evergreen |
14:45 |
Dyrcona |
stephengwills: It's looking like it might be a problem only with the Pg 10 prerequisites. |
14:49 |
Dyrcona |
It looks like it should work on Ubuntu both 18 and 20 for Pg 9.6 and Pg 10. Debian definitely doesn't install the new CPAN modules with the database prerequisites. |
14:49 |
Dyrcona |
stephengwills: I wonder if you had the right source when you did the command on the db server? Did you use git or a tarball? |
15:01 |
berick |
friday fun, i just created a VM with hostname "thecloud" |
15:04 |
Dyrcona |
:) |
15:05 |
Dyrcona |
I've named mine for distro codenames today: stretch, buster, bionic, focal. |
15:23 |
Dyrcona |
Updated the bug description to mention that it only happens on Debian: Lp 1936662. |
15:23 |
pinesol |
Launchpad bug 1936662 in Evergreen 3.7 "Missing Did You Mean Prerequisites in Pg Server Installation Make Targets" [Undecided,New] https://launchpad.net/bugs/1936662 |
15:40 |
|
jvwoolf left #evergreen |
15:41 |
Dyrcona |
Oh! They're mising in bionic, too. |
15:43 |
jeff |
well that's good, because my brain's been trying to come up with a scenario where stephengwills would have ended up with a shell where cpan was installing modules to the "wrong" shared perl libs directory, etc. |
15:46 |
Dyrcona |
jeff: That's very easy to do depending on circumstances. |
15:46 |
Dyrcona |
An environment variable, wrong user (i.e. not root/su/sudo)... |
15:47 |
Dyrcona |
i often install CPAN libs into ${HOME}/perl5/ on my workstations/laptops. |
15:56 |
* Dyrcona |
dares to run 3 vms on his laptop! |
15:56 |
Dyrcona |
Simultaneously! |
15:57 |
mmorgan |
Oooh. Dyrcona lives dangerously :) |
15:59 |
Dyrcona |
And, the patch works! |
16:00 |
Dyrcona |
I should update the commit message with verification instructions and then force push it. |
16:00 |
Dyrcona |
Still have to test on stretch and bionic, but I'm 99.9% certain that it will work there as well, since it's all the same code. |
16:15 |
|
mmorgan1 joined #evergreen |
17:21 |
|
mmorgan left #evergreen |
18:00 |
pinesol |
News from qatests: Testing Success <http://testing.evergreen-ils.org/~live> |