Summary: Fixes T10941. This avoids a confusing dead end when configuring Subversion hosting, where `svnserve` will fail to execute hooks if the CWD isn't readable by the vcs-user.
Test Plan:
- Updated and committed in a hosted SVN repository.
- Ran some git operations, too.
- @dpotter confirmed this locally in T10941.
Reviewers: chad
Reviewed By: chad
Subscribers: dpotter
Maniphest Tasks: T10941
Differential Revision: https://secure.phabricator.com/D15879
Summary:
Fixes T7475. If you do something like:
$ umask 123
$ ./bin/phd start
...the daemons might inherit the weird umask, do a `git fetch` with the weird umask, and end up creating files with weird permissions in repositories.
Instead, just normalize the umask to 022 in all cases. This is overwhelmingly the most common setting, and the one we assume things are configured with.
(When we want to force permissions to a certain setting, we do so explicitly.)
Test Plan:
- Added `var_dump(umask())` to observe umask.
- Ran `bin/phd`, saw proper umask (`18`, which is decimal of `022` octal).
- Set `umask 123`, then ran `bin/phd`, saw it correct properly again.
Reviewers: chad
Reviewed By: chad
Maniphest Tasks: T7475
Differential Revision: https://secure.phabricator.com/D15721
Summary:
Ref T4571. When a database goes down briefly, we fall back to replicas.
However, this fallback is slow (not good for users) and keeps sending a lot of traffic to the master (might be bad if the root cause is load-related).
Keep track of recent connections and fully degrade into "severed" mode if we see a sequence of failures over a reasonable period of time. In this mode, we send much less traffic to the master (faster for users; less load for the database).
We do send a little bit of traffic still, and if the master recovers we'll recover back into normal mode seeing several connections in a row succeed.
This is similar to what most load balancers do when pulling web servers in and out of pools.
For now, the specific numbers are:
- We do at most one health check every 3 seconds.
- If 5 checks in a row fail or succeed, we sever or un-sever the database (so it takes about 15 seconds to switch modes).
- If the database is currently marked unhealthy, we reduce timeouts and retries when connecting to it.
Test Plan:
- Configured a bad `master`.
- Browsed around for a bit, initially saw "unrechable master" errors.
- After about 15 seconds, saw "major interruption" errors instead.
- Fixed the config for `master`.
- Browsed around for a while longer.
- After about 15 seconds, things recovered.
- Used "Cluster Databases" console to keep an eye on health checks: it now shows how many recent health checks were good:
{F1213397}
Reviewers: chad
Reviewed By: chad
Maniphest Tasks: T4571
Differential Revision: https://secure.phabricator.com/D15677
Summary:
Ref T4571. If we fail to connect to the master, automatically try to degrade into a temporary read-only mode ("UNREACHABLE") for the remainder of the request, if possible.
If the request was something like "load the homepage", that'll work fine. If it was something like "submit a comment", there's nothing we can do and we just have to fail.
Detecting this condition imposes a performance penalty: every request checks the connection and gives the database a long time to respond, since we don't want to drop writes unless we have to. So the degraded mode works, but it's really slow, and may perpetuate the problem if the root issue is load-related.
This lays the groundwork for improving this case by degrading futher into a "SEVERED" mode which will persist across requests. In the future, if several requests in a short period of time fail, we'll sever the database host and refuse to try to connect to it for a little while, connecting directly to replicas instead (basically, we're "health checking" the master, like a load balancer would health check a web application server). This will give us a better (much faster) degraded mode in a major service disruption, and reduce load on the master if the root cause is load-related, giving it a better chance of recovering on its own.
Test Plan:
- Disabled master in config by changing the host/username, got degraded automatically to UNREACAHBLE mode immediately.
- Faked full SEVERED mode, requests hit replicas and put me in the mode properly.
- Made stuff work, hit some good pages.
- Hit some non-cluster pages.
Reviewers: chad
Reviewed By: chad
Maniphest Tasks: T4571
Differential Revision: https://secure.phabricator.com/D15674
Summary: Ref T4571. If `cluster.databases` is configured but only has replicas, implicitly drop to read-only mode and send writes to a replica.
Test Plan:
- Disabled the `master`, saw Phabricator automatically degrade into read-only mode against replicas.
- (Also tested: explicit read-only mode, non-cluster mode, properly configured cluster mode).
Reviewers: chad
Reviewed By: chad
Maniphest Tasks: T4571
Differential Revision: https://secure.phabricator.com/D15672
Summary:
Ref T4571. Allows users to click the "read-only mode" notification to get more information about why an install is in read-only mode.
Installs can be in this mode for several reasons (explicit administrative action, no masters defined, no masters reachable), and it's useful to be able to tell the difference.
Test Plan: {F1212930}
Reviewers: chad
Reviewed By: chad
Maniphest Tasks: T4571
Differential Revision: https://secure.phabricator.com/D15671
Summary:
Ref T4571. There will be a very long path beyond this, but add a basic read-only mode. You can explicitly enable this to put Phabricator in a sort of "maintenance" mode today if you're swapping databases or something.
In the long term, we'll automatically degrade into this mode if the master database is down.
Test Plan:
- Enabled read-only mode.
- Browsed around.
- Didn't immediately see anything that was totally 100% broken.
Most stuff is 80-90% broken right now. For example:
- Stuff like submitting comments doesn't work, and gives you a confusing, unhelpful error.
- None of the UI really knows that it's read-only. EditEngine stuff should all hide itself and say "you can't add new comments while an install is in read-only mode", for example, but currently does not.
Reviewers: chad
Reviewed By: chad
Maniphest Tasks: T4571
Differential Revision: https://secure.phabricator.com/D15662
Summary: Ref T7303. This application is currently stone-age tech (no transactions, hard "delete" action). Bring it up to modern specs.
Test Plan:
- Created and edited an OAuth application.
- Viewed transaction record.
- Tried to create something with no name, invalid redirect URI, etc. Was gently rebuffed with detailed explanatory errors.
Reviewers: chad
Reviewed By: chad
Maniphest Tasks: T7303
Differential Revision: https://secure.phabricator.com/D15609
Summary: Ref T7053. Remove the `envHash` and `envInfo` fields, which are no longer used now that the daemons restart automagically. Depends on D14458.
Test Plan: Saw no more setup issues.
Reviewers: #blessed_reviewers, epriestley
Reviewed By: #blessed_reviewers, epriestley
Subscribers: tycho.tatitscheff, epriestley
Maniphest Tasks: T7053
Differential Revision: https://secure.phabricator.com/D14446
Summary: Use `PhutilClassMaQuery` instead of `PhutilSymbolLoader`, mostly for consistency. Depends on D13588.
Test Plan: Poked around a bunch of pages.
Reviewers: #blessed_reviewers, epriestley
Reviewed By: #blessed_reviewers, epriestley
Subscribers: epriestley, Korvin
Differential Revision: https://secure.phabricator.com/D13589
Summary: All classes should extend from some other class. See D13275 for some explanation.
Test Plan: `arc unit`
Reviewers: epriestley, #blessed_reviewers
Reviewed By: epriestley, #blessed_reviewers
Subscribers: epriestley, Korvin
Differential Revision: https://secure.phabricator.com/D13283
Summary: Ref T6367.
Test Plan:
- Added and executed unit tests.
- Sent mail to A (en_US) and B (en_A*).
- Got one mail in English and one mail in ENGLISH.
Reviewers: btrahan
Reviewed By: btrahan
Subscribers: epriestley
Maniphest Tasks: T6367
Differential Revision: https://secure.phabricator.com/D13142
Summary: Use `__CLASS__` instead of hard-coding class names. Depends on D12605.
Test Plan: Eyeball it.
Reviewers: #blessed_reviewers, epriestley
Reviewed By: #blessed_reviewers, epriestley
Subscribers: hach-que, Korvin, epriestley
Differential Revision: https://secure.phabricator.com/D12806
Summary: Ref T6755. I'll add some notes there about specifics.
Test Plan:
- Made connections to HTTP and HTTPS URIs.
- Added some debugging code to verify that HTTP URIs were pre-resolved.
Reviewers: btrahan
Reviewed By: btrahan
Subscribers: epriestley
Maniphest Tasks: T6755
Differential Revision: https://secure.phabricator.com/D12169
Summary:
Ref T6755. This is a partial fix, but:
- Allow netblocks to be blacklisted instead of making the feature all-or-nothing.
- Default to disallow requests to all reserved private/local/special IP blocks. This should generally be a "safe" setting.
- Explain the risks better.
- Improve the errors rasied by Macro when failing.
- Removed `security.allow-outbound-http`, as it is superseded by this setting and is somewhat misleading.
- We still make outbound HTTP requests to OAuth.
- We still make outbound HTTP requests for repositories.
From a technical perspective:
- Separate URIs that are safe to link to or redirect to (basically, not "javascript://") from URIs that are safe to fetch (nothing in a private block).
- Add the default blacklist.
- Be more careful with response data in Macro fetching, and don't let the user see it if it isn't ultimately valid.
Additionally:
- I want to do this check before pulling repositories, but that's enough of a mess that it should go in a separate diff.
- The future implementation of T4190 needs to perform the fetch check.
Test Plan:
- Fetched a valid macro.
- Fetched a non-image, verified it didn't result in a viewable file.
- Fetched a private-ip-space image, got an error.
- Fetched a 404, got a useful-enough error without additional revealing response content (which is usually HTML anyway and not useful).
- Fetched a bad protocol, got an error.
- Linked to a local resource, a phriction page, a valid remote site, all worked.
- Linked to private IP space, which worked fine (we want to let you link and redierect to other private services, just not fetch them).
- Added and executed unit tests.
Reviewers: btrahan
Reviewed By: btrahan
Subscribers: epriestley
Maniphest Tasks: T6755
Differential Revision: https://secure.phabricator.com/D12136
Summary:
Ref T7152. Ref T1139. This updates Phabricator so third-party libraries can translate their own stuff. Also:
- Hide "All Caps" when not in development mode, since some users have found this a little confusing.
- With other changes, adds a "Raw Strings" mode (development mode only).
- Add an example silly translation to make sure the serious business flag works.
- Add a basic British English translation.
- Simplify handling of translation overrides.
Test Plan:
- Flipped serious business / development on and off and saw silly/development translations drop off.
- Switched to "All Caps" and saw all caps.
- Switched to Very English, Wow!
- Switched to British english and saw "colour".
Reviewers: btrahan
Reviewed By: btrahan
Subscribers: epriestley
Maniphest Tasks: T7152, T1139
Differential Revision: https://secure.phabricator.com/D11747
Summary:
I'm hitting this in the cluster and couldn't figure it out after staring at it for a couple minutes. Produce a better error.
This dumps a hash of each configuration key value which is set to a non-default value into the daemon log. This is much more compact than the full config, and doesn't spread secrets around, so it seems like a good balance between providing information and going crazy with it.
Test Plan: {F284139}
Reviewers: btrahan
Reviewed By: btrahan
Subscribers: epriestley
Differential Revision: https://secure.phabricator.com/D11699
Summary:
Fixes T7019. In a cluster environment, pushes currently fail because the commit hook can't identify the instance.
For web processes, the hostname identifies the instance -- but we don't have a hostname in the hook.
For CLI processes, the environment identifies the instance -- but we don't have an environment in the hook under SVN.
Promote the instance identifier into the upstream and pack/unpack it explicitly for hooks. This is probably not useful for anyone but us, but the amount of special-purpose code we're introducing is very small.
I poked at trying to do this in a more general way, but:
- We MUST know this BEFORE we run code, so the normal subclassing stuff is useless.
- I couldn't come up with any other parameter which might ever be useful to pass in.
Test Plan: Used `git push` to push code through proxied HTTP, got a clean push.
Reviewers: btrahan
Reviewed By: btrahan
Subscribers: epriestley
Maniphest Tasks: T7019
Differential Revision: https://secure.phabricator.com/D11495
Summary: Fixes T6959.
Test Plan: When I was ready to test the feature, the "Daemon & Web config" error already showed up, from having added phd.variant-config. I went meta and changed the value of phd.variant-config to have phd.variant-config. The config error disappeared. I then changed the conpherence setting about conpherence email prefix and the error showed up again. Removing the conpherence config setting made the error disappear once more.
Reviewers: epriestley
Reviewed By: epriestley
Subscribers: Korvin, epriestley
Maniphest Tasks: T6959
Differential Revision: https://secure.phabricator.com/D11399
Summary:
Ref T2783. Ref T6706.
- Add `cluster.addresses`. This is a whitelist of CIDR blocks which define cluster hosts.
- When we recieve a request that has a cluster-based authentication token, require the cluster to be configured and require the remote address to be a cluster member before we accept it.
- This provides a general layer of security for these mechanisms.
- In particular, it means they do not work by default on unconfigured hosts.
- When cluster addresses are configured, and we receive a request //to// an address not on the list, reject it.
- This provides a general layer of security for getting the Ops side of cluster configuration correct.
- If cluster nodes have public IPs and are listening on them, we'll reject requests.
- Basically, this means that any requests which bypass the LB get rejected.
Test Plan:
- With addresses not configured, tried to make requests; rejected for using a cluster auth mechanism.
- With addresses configred wrong, tried to make requests; rejected for sending from (or to) an address outside of the cluster.
- With addresses configured correctly, made valid requests.
Reviewers: btrahan
Reviewed By: btrahan
Subscribers: epriestley
Maniphest Tasks: T6706, T2783
Differential Revision: https://secure.phabricator.com/D11159
Summary:
Fixes T2792. This adds a pluggable configuration layer between all the stuff on disk (local/file) and the runtime configurable stuff (database).
An install can subclass this source and:
- For Phacility, query a remote service (like Almanac) to retrieve hostname-based configuration, allowing one install to serve multiple instances.
- Maybe for Phacility, query a remote service (like Phlux) to retrieve sitevar-like configuration (e.g., put everything in readonly mode to deal with a maintenance issue?). Not sure if we'll do this or not. We might just nuke Phlux since Almanac is sort-of-a-superset of it for our purposes.
- For third parties, query some other remote service if that makes config management easier. In particular, it would theoretically let you put locked config in Zookeeper or whatever else you want.
Test Plan: Added a fake source and saw it inject configuration.
Reviewers: btrahan
Reviewed By: btrahan
Subscribers: epriestley
Maniphest Tasks: T2792
Differential Revision: https://secure.phabricator.com/D10787
Summary: Shows the UI everywhere. Also asort() the keys before calculating the environment hash as that is probably an issue for someone at some point we just don't need to have. Ref T5968.
Test Plan: Viewed the setup check and saw a link to the daemon console. Viewed the daemon console and saw the various stale config daemons. Clicked a daemon and saw a "stale config" header icon where expected. Restarted daemons and all of this went away.
Reviewers: epriestley
Reviewed By: epriestley
Subscribers: epriestley, Korvin
Maniphest Tasks: T5968
Differential Revision: https://secure.phabricator.com/D10367
Summary: Fixes T4881.
Test Plan: made a config change, saw the issue, restarted daemons and it went away
Reviewers: epriestley
Reviewed By: epriestley
Subscribers: epriestley, Korvin
Maniphest Tasks: T4881
Differential Revision: https://secure.phabricator.com/D10339
Summary:
Via HackerOne. Chrome (at least) interprets backslashes like forward slashes, so a redirect to "/\evil.com" is the same as a redirect to "//evil.com".
- Reject local URIs with backslashes (we never generate these).
- Fully-qualify all "Location:" redirects.
- Require external redirects to be marked explicitly.
Test Plan:
- Expanded existing test coverage.
- Verified that neither Diffusion nor Phriction can generate URIs with backslashes (they are escaped in Diffusion, and removed by slugging in Phriction).
- Logged in with Facebook (OAuth2 submits a form to the external site, and isn't affected) and Twitter (OAuth1 redirects, and is affected).
- Went through some local redirects (login, save-an-object).
- Verified file still work.
Reviewers: btrahan
Reviewed By: btrahan
Subscribers: epriestley
Differential Revision: https://secure.phabricator.com/D10291
Summary: Ran `arc lint --apply-patches --everything` over rP, mainly to change double quotes to single quotes where appropriate. These changes also validate that the `ArcanistXHPASTLinter::LINT_DOUBLE_QUOTE` rule is working as expected.
Test Plan: Eyeballed it.
Reviewers: #blessed_reviewers, epriestley
Reviewed By: #blessed_reviewers, epriestley
Subscribers: epriestley, Korvin, hach-que
Differential Revision: https://secure.phabricator.com/D9431
Summary:
- Point them at the new Diviner.
- Make them a little less cumbersome to write.
Test Plan: Found almost all of these links in the UI and clicked them.
Reviewers: btrahan
Reviewed By: btrahan
Subscribers: epriestley
Differential Revision: https://secure.phabricator.com/D8553
Summary:
There are quite a few tests in Arcanist, libphutil and Phabricator that do something similar to `$this->assertEqual(false, ...)` or `$this->assertEqual(true, ...)`.
This is unnecessarily verbose and it would be cleaner if we had `assertFalse` and `assertTrue` methods.
Test Plan: I contemplated adding a unit test for the `getCallerInfo` method but wasn't sure if it was required / where it should live.
Reviewers: epriestley, #blessed_reviewers
Reviewed By: epriestley
CC: Korvin, epriestley, aran
Differential Revision: https://secure.phabricator.com/D8460
Summary:
Ref T603. I had to partially revert this earlier because it accidentally blocked access to Conduit and File data for installs without "policy.allow-public", since the applications are available to "all users" but some endpoints actually need to be available even when not logged in.
This readjusts the gating in the controller to properly apply application visibility restrictions, and then adds a giant pile of unit test coverage to make sure it sticks and all the weird cases are covered.
Test Plan:
- Added and executed unit tests.
- Executed most of the tests manually, by using logged in / admin / public / disabled users.
Reviewers: btrahan
Reviewed By: btrahan
CC: aran
Maniphest Tasks: T603
Differential Revision: https://secure.phabricator.com/D7211
Summary: Ref T1536. This is the last major migration. Moves us over to the DB and drops all the config stuff.
Test Plan:
- Ran the migration.
- Saw all my old config brought forward and respected, with accurate settings.
- Ran LDAP import.
- Grepped for all removed config options.
Reviewers: btrahan, chad
Reviewed By: btrahan
CC: aran, wez
Maniphest Tasks: T1536
Differential Revision: https://secure.phabricator.com/D6243
Summary:
Kind of a quick look at an idea for T2184
Ref T2184
Test Plan: Make sure the site still loads
Reviewers: epriestley
CC: aran, Korvin, mbishopim3
Maniphest Tasks: T2184
Differential Revision: https://secure.phabricator.com/D6045
Summary:
It's sometimes necessary to specify the paths to individual binaries
explicitly, e.g. a particular build of 'javelinsymbols' or a newer
version of git than is installed on your shared system.
By adding symlinks in the .../phabricator/support/bin/ directory you
can now spell these out using the file system.
Test Plan:
Ran on local Ubuntu VM:
.. add 'TEST' repo to diffusion ..
.. visit 127.0.01/diffusion/TEST - see ok ..
$ cd /opt
$ sudo sh -c 'echo "exit 1" > badgit'
$ sudo chmod +x /opt/badgit
$ sudo mkdir goodgit
$ sudo mv /usr/bin/git /opt/goodgit/
.. unset environment.append-paths ..
.. visit 127.0.01/diffusion/TEST - see error 'git: not found' ..
.. set environment.append-paths to /opt/goodgit/ ..
.. visit 127.0.01/diffusion/TEST - see ok ..
$ sudo ln -s /opt/badgit /usr/bin/git
.. visit 127.0.01/diffusion/TEST - see error 'error #1' ..
sudo ln -s /opt/goodgit/git web/phabricator/support/bin/git
.. visit 127.0.01/diffusion/TEST - see ok ..
.. unset environment.append-paths ..
.. visit 127.0.01/diffusion/TEST - see ok ..
$ sudo rm web/phabricator/support/bin/git
.. visit 127.0.01/diffusion/TEST - see error 'error #1' ..
$ sudo rm /usr/bin/git
$ sudo mv /opt/goodgit/git /usr/bin/
.. visit 127.0.01/diffusion/TEST - see ok ..
Note that 'DIRECTORY_SEPARATOR' was not used because apparently it's
portable and ok to just use '/'.
http://alanhogan.com/tips/php/directory-separator-not-necessary
(I'm pretty new to PHP so looking for guidance :)
Reviewers: epriestley
Reviewed By: epriestley
CC: aran, Korvin
Maniphest Tasks: T2378
Differential Revision: https://secure.phabricator.com/D5561
Summary: We hit some env config fairly often (~200 calls on the home page) and it's not especially cheap right now. This saves about 10ms on home, which is ~8% of the page weight on my machine.
Test Plan: Clicked around, everything seemed fine. Unit tests.
Reviewers: btrahan
Reviewed By: btrahan
CC: aran
Differential Revision: https://secure.phabricator.com/D5512
Summary:
Fixes T2059. Ref T2517.
Currently, you can run `bin/storage upgrade` with `--user` and `--password` arguments. However, these clownishly apply only to `.sql` patches -- the `.php` migrations still use the default user and password.
This is dumb. Stop doing it. Respect `--user` and `--password` for PHP patches.
(I implemented "override", which is very similar to "repair", but kept them separate since I think they're semantically distinct enough to differentiate.)
Test Plan: Ran `./bin/storage upgrade --user x --pass y --apply phabricator:20130219.commitsummarymig.php`. Verified the correct user and password were used both for the initial connect and patch application.
Reviewers: chad, vrana
Reviewed By: chad
CC: aran
Maniphest Tasks: T2059, T2517
Differential Revision: https://secure.phabricator.com/D5115
Summary:
- PHABRICATOR_ENV is now optional. If you don't specify it, we won't load a config file.
- PhabricatorSetup is now gone.
- I removed the alternate file domain check for now, see T2380.
- `phabricator.setup` config is now gone.
- Rewrote documentation:
- No more mentions of `phabricator.setup`.
- Normal install guide no longer mentions PHABRICATOR_ENV. This is now an advanced topic.
- Clarified that you only need to set up one of apache, nginx or lighttpd.
- Tweaked a few things I've seen users have difficulty with.
This should have no effect on any existing installs, but make the process much simpler for future installs.
Closes T2221.
Closes T2223.
Closes T2228.
Test Plan:
- Removed my PHABRICATOR_ENV and went through the install process.
- Generated and read documentation.
Reviewers: btrahan, chad
Reviewed By: btrahan
CC: aran
Maniphest Tasks: T2221, T2223, T2228
Differential Revision: https://secure.phabricator.com/D4596
Summary:
Fixes T2293.
We currently hard-require this in setup. We do not need to; we don't actually need it until we start running daemons. Move it to post-install and provide more guidance.
We could make this even easier in the future, but we'd need to special case it, since it's dangerous to let it be set to any value (if you set it to the wrong value, you can't log in). We could safely have a workflow which writes the current request URI into the database configuration, or a two-stage workflow where we set the URI and then verify it, but these both imply some special casing and complication. This should be a step forward from where we are today, regardless.
Test Plan:
Removed "phabricator.base-uri" from my configuration. Verified Phabricator still works.
Without "phabricator.base-uri" configured, logged in from multiple host names (127.0.0.1:8080, local.aphront.com:8080).
Configured "phabricator.base-uri". Verified my unblessed session no longer worked. Verified setup issue went away.
Reviewers: btrahan, vrana
Reviewed By: btrahan
CC: aran
Maniphest Tasks: T2293
Differential Revision: https://secure.phabricator.com/D4580
Summary: Unit test for T2345
Test Plan: Ran unit tests, checked that it passed.
Reviewers: epriestley
Reviewed By: epriestley
CC: aran, Korvin
Differential Revision: https://secure.phabricator.com/D4552
Summary:
T2345
getConfig throws an Exception when the key does not exist.
Also removes dead code that throws an Exception.
Test Plan:
Reloaded the Phabricator home page. In the process, found
2 Exceptions thrown due to nonexistent keys. After addressing these problems,
the home page loads without Exceptions.
Reviewers: epriestley
Reviewed By: epriestley
CC: aran, Korvin
Differential Revision: https://secure.phabricator.com/D4541
Summary: Fixes T2269. If the user manages to mess up both the PHP and Phabricator configurations, set the timezone to UTC. We basically never use this anyway (we always render into the user's time), PHP just gets angry at us if we don't set it. (We do use it for logged-out users, I suppose.)
Test Plan: Set PHP and Phabricator timezones to goofy nonsense, verified we recover sensibly from it.
Reviewers: btrahan, vrana
Reviewed By: btrahan
CC: aran
Maniphest Tasks: T2228, T2269
Differential Revision: https://secure.phabricator.com/D4496
Summary: Because the Default configuration provider is loaded before custom libraries, any config options specified in them don't get a default values.
Test Plan: Looked at /config/
Reviewers: epriestley, codeblock, btrahan
Reviewed By: epriestley
CC: aran, Korvin
Differential Revision: https://secure.phabricator.com/D4532
Summary: This is gross, but fixes an issue where `bin/storage upgrade` tries to access DB config which doesn't exist yet. We need a version of this for `bin/config` anyway. I'll sort this out into a proper sequenced startup process in a followup.
Test Plan: `bin/storage upgrade` no longer fatals when upgrading across the config boundary.
Reviewers: asherkin, codeblock, btrahan, vrana
Reviewed By: codeblock
CC: aran
Differential Revision: https://secure.phabricator.com/D4512
Summary:
If your configuration overrides the connection adapter, we need to load libraries before we can setup the database config source.
Also lock this since it won't work when edited from the web anymore, and so sneaky users can't upload stuff and then edit their config to run arbitrary code.
Test Plan: See chatlog in #phabricator. This is a problem for Facebook only.
Reviewers: vrana, btrahan
Reviewed By: vrana
CC: aran
Differential Revision: https://secure.phabricator.com/D4498
Summary:
When configuration is set incorrectly (e.g., of the wrong type), detect and repair it by setting it to the default value. A setup warning will be raised separately.
Notably, this removes the need to hard-code all the class types.
This runs separately from the "invalid config" check because we need to run it on every page, but do setup checks only once per restart (some of them are slow).
Also dirty setup when we edit configuration.
Test Plan: Set config incorrectly on purpose, saw Phabricator correct it on restart and on every subsequent page load until it was fixed.
Reviewers: btrahan, vrana
Reviewed By: vrana
CC: aran
Maniphest Tasks: T2292
Differential Revision: https://secure.phabricator.com/D4492
Summary:
Read configuration from the new database source.
This adds an extra MySQL connect + query to every page. They're very cheap so I think we can suffer them for now, but I'd like to put cache in front of this at some point. The difficulties are:
- If we use APC, multi-frontend installs (Facebook) can't dirty it (major problem), and the CLI can't dirty it (fine for now, maybe a major problem later).
- If we use Memcache, we need to add config stuff.
- We could use APC in all non-Facebook installs if we can make it dirtyable from the CLI, but I don't see a reasonable way to do that.
- We don't have any other caches which are faster than the database.
So I'll probably implement Memcache support at some point, although this is a lame excuse for it.
Test Plan: Added some config values via web UI, saw them active on the install.
Reviewers: btrahan, codeblock, vrana
Reviewed By: codeblock
CC: aran
Maniphest Tasks: T2221
Differential Revision: https://secure.phabricator.com/D4296
