04cf68b45e make nginx responsible for downloading (and caching)
files from S3. As noted in that commit, nginx implements its own
non-blocking DNS resolver, since the base syscall is blocking, so
requires an explicit nameserver configuration. That commit used
127.0.0.53, which is provided by systemd-resolved, as the resolver.
However, that service may not always be enabled and running, and may
in fact not even be installed (e.g. on Docker). Switch to parsing
`/etc/resolv.conf` and using the first-provided nameserver. In many
deployments, this will still be `127.0.0.53`, but for others it will
provide a working DNS server which is external to the host.
In the event that a server is misconfigured and has no resolvers in
`/etc/resolv.conf`, it will error out:
```console
Error: Evaluation Error: Error while evaluating a Function Call, No nameservers found in /etc/resolv.conf! Configure one by setting application_server.nameserver in /etc/zulip/zulip.conf (file: /home/zulip/deployments/current/puppet/zulip/manifests/app_frontend_base.pp, line: 76, column: 70) on node example.zulipdev.org
```
Django 4.0 and higher began checking the `Origin` header, which made
it important that Zulip know accurately if the request came over HTTPS
or HTTP; failure to do so would result in "CSRF verification failed"
errors.
For Zulip servers which are accessed via proxies, this means that
`X-Fowarded-Proto` must be set accurately. Adjust the documentation
for the suggested configurations to add the header.
Fixes: #24599.
Co-authored-by: Alex Vandiver <alexmv@zulip.com>
1c76036c61 raised the number of `minfds` in Supervisor from 40k to
1M. If Supervisor cannot guarantee that number of available file
descriptors, it will fail to start; `/etc/security/limits.conf` was
hence adjusted upwards as well. However, on some virtualized
environments, including Proxmox LXC, setting
`/etc/security/limits.conf` may not be enough to raise the
system-level limits. This causes `supervisord` with the larger
`minfds` to fail to start.
The limit of 1000000 was chosen to be arbitrarily high, assuming it
came without cost; it is not expected to ever be reached on any
deployment. 262b19346e already lowered one aspect of that
changeset, upon determining it did come with a cost. Potentially
breaking virtualized deployments during upgrade is another cost of
that change.
Lower the `minfds` it back down to 40k, partially reverting
1c76036c61, but allow adjusting it upwards for extremely large
deployments. We do not expect any except the largest deployments to
ever hit the 40k limit, and a frictionless deployment for the
vanishingly small number of huge deployments is not worth the
potential upgrade hiccups for the much more frequent smaller
deployments.
Instead of copying over a mostly-unchanged `postgresql.conf`, we
transition to deploying a `conf.d/zulip.conf` which contains the
only material changes we made to the file, which were previously
appended to the end.
While shipping separate while `postgresql.conf` files for each
supported version is useful if there is large variety in supported
options between versions, there is not no such variation at current,
and the burden of overriding the entire default configuration is that
it must be keep up to date wit the package's version.
Taking backups on the database primary adds additional disk load,
which can impact the performance of the application.
Switch to taking backups on replicas, if they exist. Some deployments
may have multiple replicas, and taking backups on all of them is
wasteful and potentially confusing; add a flag to inhibit taking
nightly snapshots on the host.
If the deployment is a single instance of PostgreSQL, with no
replicas, it takes backups as before, modulo the extra flag to allow
skipping taking them.
Since logrotate runs in a daily cron, this practically means "daily,
but only if it's larger than 500M." For large installs with large
traffic, this is effectively daily for 10 days; for small installs, it
is an unknown amount of time.
Switch to daily logfiles, defaulting to 14 days to match nginx; this
can be overridden using a zulip.conf setting. This makes it easier to
ensure that access logs are only kept for a bounded period of time.
Increasing worker_connections has a memory cost, unlike the rest of
the changes in 1c76036c61d8; setting it to 1 million caused nginx to
consume several GB of memory.
Reduce the default down to 10k, and allow deploys to configure it up
if necessary. `worker_rlimit_nofile` is left at 1M, since it has no
impact on memory consumption.
These hooks are run immediately around the critical section of the
upgrade. If the upgrade fails for preparatory reasons, the pre-deploy
hook may not be run; if it fails during the upgrade, the post-deploy
hook will not be run. Hooks are called from the CWD of the new
deploy, with arguments of the old version and the new version. If
they exit with non-0 exit code, the deploy aborts.
- Renames "Customize Zulip" to "Server configuration".
- Cross-links "Server configuration" with "System and deployment
configuration".
Fixes part of #23984.
The `postfix.mailname` setting in `/etc/zulip.conf` was previously
only used for incoming mail, to identify in Postfix configuration
which messages were "local."
Also set `/etc/mailname`, which is used by Postfix to set how it
identifies to other hosts when sending outgoing email.
Co-authored-by: Alex Vandiver <alexmv@zulip.com>
When file uploads are stored in S3, this means that Zulip serves as a
302 to S3. Because browsers do not cache redirects, this means that
no image contents can be cached -- and upon every page load or reload,
every recently-posted image must be re-fetched. This incurs extra
load on the Zulip server, as well as potentially excessive bandwidth
usage from S3, and on the client's connection.
Switch to fetching the content from S3 in nginx, and serving the
content from nginx. These have `Cache-control: private, immutable`
headers set on the response, allowing browsers to cache them locally.
Because nginx fetching from S3 can be slow, and requests for uploads
will generally be bunched around when a message containing them are
first posted, we instruct nginx to cache the contents locally. This
is safe because uploaded file contents are immutable; access control
is still mediated by Django. The nginx cache key is the URL without
query parameters, as those parameters include a time-limited signed
authentication parameter which lets nginx fetch the non-public file.
This adds a number of nginx-level configuration parameters to control
the caching which nginx performs, including the amount of in-memory
index for he cache, the maximum storage of the cache on disk, and how
long data is retained in the cache. The currently-chosen figures are
reasonable for small to medium deployments.
The most notable effect of this change is in allowing browsers to
cache uploaded image content; however, while there will be many fewer
requests, it also has an improvement on request latency. The
following tests were done with a non-AWS client in SFO, a server and
S3 storage in us-east-1, and with 100 requests after 10 requests of
warm-up (to fill the nginx cache). The mean and standard deviation
are shown.
| | Redirect to S3 | Caching proxy, hot | Caching proxy, cold |
| ----------------- | ------------------- | ------------------- | ------------------- |
| Time in Django | 263.0 ms ± 28.3 ms | 258.0 ms ± 12.3 ms | 258.0 ms ± 12.3 ms |
| Small file (842b) | 586.1 ms ± 21.1 ms | 266.1 ms ± 67.4 ms | 288.6 ms ± 17.7 ms |
| Large file (660k) | 959.6 ms ± 137.9 ms | 609.5 ms ± 13.0 ms | 648.1 ms ± 43.2 ms |
The hot-cache performance is faster for both large and small files,
since it saves the client the time having to make a second request to
a separate host. This performance improvement remains at least 100ms
even if the client is on the same coast as the server.
Cold nginx caches are only slightly slower than hot caches, because
VPC access to S3 endpoints is extremely fast (assuming it is in the
same region as the host), and nginx can pool connections to S3 and
reuse them.
However, all of the 648ms taken to serve a cold-cache large file is
occupied in nginx, as opposed to the only 263ms which was spent in
nginx when using redirects to S3. This means that to overall spend
less time responding to uploaded-file requests in nginx, clients will
need to find files in their local cache, and skip making an
uploaded-file request, at least 60% of the time. Modeling shows a
reduction in the number of client requests by about 70% - 80%.
The `Content-Disposition` header logic can now also be entirely shared
with the local-file codepath, as can the `url_only` path used by
mobile clients. While we could provide the direct-to-S3 temporary
signed URL to mobile clients, we choose to provide the
served-from-Zulip signed URL, to better control caching headers on it,
and greater consistency. In doing so, we adjust the salt used for the
URL; since these URLs are only valid for 60s, the effect of this salt
change is minimal.
Moving `/user_avatars/` to being served partially through Django
removes the need for the `no_serve_uploads` nginx reconfiguring when
switching between S3 and local backends. This is important because a
subsequent commit will move S3 attachments to being served through
nginx, which would make `no_serve_uploads` entirely nonsensical of a
name.
Serve the files through Django, with an offload for the actual image
response to an internal nginx route. In development, serve the files
directly in Django.
We do _not_ mark the contents as immutable for caching purposes, since
the path for avatar images is hashed only by their user-id and a salt,
and as such are reused when a user's avatar is updated.
The default value in uwsgi is 4k; receiving more than this amount from
nginx leads to a 502 response (though, happily, the backend uwsgi does not
terminate).
ab18dbfde5 originally increased it from the unstated uwsgi default
of 4096, to 8192; b1da797955 made it configurable, in order to allow
requests from clients with many cookies, without causing 502's[1].
nginx defaults to a limitation of 1k, with 4 additional 8k header
lines allowed[2]; any request larger than that returns a response of
`400 Request Header Or Cookie Too Large`. The largest header size
theoretically possible from nginx, by default, is thus 33k, though
that would require packing four separate headers to exactly 8k each.
Remove the gap between nginx's limit and uwsgi's, which could trigger
502s, by removing the uwsgi configurability, and setting a 64k size in
uwsgi (the max allowable), which is larger than nginx's default limit.
uWSGI's documentation of `buffer-size` ([3], [4]) also notes that "It
is a security measure too, so adapt to your app needs instead of
maxing it out." Python has no security issues with buffers of 64k,
and there is no appreciable memory footprint difference to having a
larger buffer available in uwsgi.
[1]: https://chat.zulip.org/#narrow/stream/31-production-help/topic/works.20in.20Edge.20not.20Chrome/near/719523
[2]: https://nginx.org/en/docs/http/ngx_http_core_module.html#client_header_buffer_size
[3]: https://uwsgi-docs.readthedocs.io/en/latest/ThingsToKnow.html
[4]: https://uwsgi-docs.readthedocs.io/en/latest/Options.html#buffer-size
Backups are written every 16k of WAL archive, and by default do not
have an upper limit on how out of date they are, as `archive_timeout`
defaults to 0.
Also emphasize that these are streaming backups, not just one
point-in-time backup daily.
Fixes#21976.
Notable changes:
- Describe `X-Forwarded-For` by name.
- Switch each specific proxy to numbered steps.
- Link back to the `X-Forwarded-For` section in each proxy
- Default to using HTTPS, not HTTP, for the backend.
- Include the HTTP-to-HTTPS redirect code for all proxies; it is
important that it happen at the proxy, as the backend is unaware of
it.
- Call out Apache2 modules which are necessary.
- Specify where the dhparam.pem file can be found.
- Call out the `Host:` header forwarding necessary, and document
`USE_X_FORWARDED_HOST` if that is not possible.
- Standardize on 20 minutes of connection timeout.
When the credentials are provided by dint of being run on an EC2
instance with an assigned Role, we must be able to fetch the instance
metadata from IMDS -- which is precisely the type of internal-IP
request that Smokescreen denies.
While botocore supports a `proxies` argument to the `Config` object,
this is not actually respected when making the IMDS queries; only the
environment variables are read from. See
https://github.com/boto/botocore/issues/2644
As such, implement S3_SKIP_PROXY by monkey-patching the
`botocore.utils.should_bypass_proxies` function, to allow requests to
IMDS to be made without Smokescreen impeding them.
Fixes#20715.
Previously, it was possible to configure `wal-g` backups without
replication enabled; this resulted in only daily backups, not
streaming backups. It was also possible to enable replication without
configuring the `wal-g` backups bucket; this simply failed to work.
Make `wal-g` backups always streaming, and warn loudly if replication
is enabled but `wal-g` is not configured.
This uses the myst_heading_anchors option to automatically generate
header anchors and make Sphinx aware of them. See
https://myst-parser.readthedocs.io/en/latest/syntax/optional.html#auto-generated-header-anchors.
Note: to be compatible with GitHub, MyST-Parser uses a slightly
different convention for .md fragment links than .html fragment links
when punctuation is involved. This does not affect the generated
fragment links in the HTML output.
Fixes#13264.
Signed-off-by: Anders Kaseorg <anders@zulip.com>
We previously had a convention of redundantly including the directory
in relative links to reduce mistakes when moving content from one file
to another. However, these days we have a broken link checker in
test-documentation, and after #21237, MyST-Parser will check relative
links (including fragments) when you run build-docs.
Signed-off-by: Anders Kaseorg <anders@zulip.com>
This is required in order to lock down the RabbitMQ port to only
listen on localhost. If the nodename is `rabbit@hostname`, in most
circumstances the hostname will resolve to an external IP, which the
rabbitmq port will not be bound to.
Installs which used `rabbit@hostname`, due to RabbitMQ having been
installed before Zulip, would not have functioned if the host or
RabbitMQ service was restarted, as the localhost restrictions in the
RabbitMQ configuration would have made rabbitmqctl (and Zulip cron
jobs that call it) unable to find the rabbitmq server.
The previous commit ensures that configure-rabbitmq is re-run after
the nodename has changed. However, rabbitmq needs to be stopped
before `rabbitmq-env.conf` is changed; we use an `onlyif` on an `exec`
to print the warning about the node change, and let the subsequent
config change and notify of the service and configure-rabbitmq to
complete the re-configuration.
Because Camo includes logic to deny access to private subnets, routing
its requests through Smokescreen is generally not necessary. However,
it may be necessary if Zulip has configured a non-Smokescreen exit
proxy.
Default Camo to using the proxy only if it is not Smokescreen, with a
new `proxy.enable_for_camo` setting to override this behaviour if need
be. Note that that setting is in `zulip.conf` on the host with Camo
installed -- not the Zulip frontend host, if they are different.
Fixes: #20550.
For `no_serve_uploads`, `http_only`, which previously specified
"non-empty" to enable, this tightens what values are true. For
`pgroonga` and `queue_workers_multiprocess`, this broadens the
possible values from `enabled`, and `true` respectively.
Restarting the uwsgi processes by way of supervisor opens a window
during which nginx 502's all responses. uwsgi has a configuration
called "chain reloading" which allows for rolling restart of the uwsgi
processes, such that only one process at once in unavailable; see
uwsgi documentation ([1]).
The tradeoff is that this requires that the uwsgi processes load the
libraries after forking, rather than before ("lazy apps"); in theory
this can lead to larger memory footprints, since they are not shared.
In practice, as Django defers much of the loading, this is not as much
of an issue. In a very basic test of memory consumption (measured by
total memory - free - caches - buffers; 6 uwsgi workers), both
immediately after restarting Django, and after requesting `/` 60 times
with 6 concurrent requests:
| Non-lazy | Lazy app | Difference
------------------+------------+------------+-------------
Fresh | 2,827,216 | 2,870,480 | +43,264
After 60 requests | 3,332,284 | 3,409,608 | +77,324
..................|............|............|.............
Difference | +505,068 | +539,128 | +34,060
That is, "lazy app" loading increased the footprint pre-requests by
43MB, and after 60 requests grew the memory footprint by 539MB, as
opposed to non-lazy loading, which grew it by 505MB. Using wsgi "lazy
app" loading does increase the memory footprint, but not by a large
percentage.
The other effect is that processes may be served by either old or new
code during the restart window. This may cause transient failures
when new frontend code talks to old backend code.
Enable chain-reloading during graceful, puppetless restarts, but only
if enabled via a zulip.conf configuration flag.
Fixes#2559.
[1]: https://uwsgi-docs.readthedocs.io/en/latest/articles/TheArtOfGracefulReloading.html#chain-reloading-lazy-apps
The certbot package installs its own systemd timer (and cron job,
which disabled itself if systemd is enabled) which updates
certificates. This process races with the cron job which Zulip
installs -- the only difference being that Zulip respects the
`certbot.auto_renew` setting, and that it passes the deploy hook.
This means that occasionally nginx would not be reloaded, when the
systemd timer caught the expiration first.
Remove the custom cron job and `certbot-maybe-renew` script, and
reconfigure certbot to always reload nginx after deploying, using
certbot directory hooks.
Since `certbot.auto_renew` can't have an effect, remove the setting.
In turn, this removes the need for `--no-zulip-conf` to
`setup-certbot`. `--deploy-hook` is similarly removed, as running
deploy hooks to restart nginx is now the default; pass
`--no-directory-hooks` in standalone mode to not attempt to reload
nginx. The other property of `--deploy-hook`, of skipping symlinking
into place, is given its own flog.
PostgreSQL 11 and below used a configuration file names
`recovery.conf` to manage replicas and standbys; support for this was
removed in PostgreSQL 12[1], and the configuration parameters were
moved into the main `postgresql.conf`.
Add `zulip.conf` settings for the primary server hostname and
replication username, so that the complete `postgresql.conf`
configuration on PostgreSQL 14 can continue to be managed, even when
replication is enabled. For consistency, also begin writing out the
`recovery.conf` for PostgreSQL 11 and below.
In PostgreSQL 12 configuration and later, the `wal_level =
hot_standby` setting is removed, as `hot_standby` is equivalent to
`replica`, which is the default value[2]. Similarly, the
`hot_standby = on` setting is also the default[3].
Documentation is added for these features, and the commentary on the
"Export and Import" page referencing files under `puppet/zulip_ops/`
is removed, as those files no longer have any replication-specific
configuration.
[1]: https://www.postgresql.org/docs/current/recovery-config.html
[2]: https://www.postgresql.org/docs/12/runtime-config-wal.html#GUC-WAL-LEVEL
[3]: https://www.postgresql.org/docs/12/runtime-config-replication.html#GUC-HOT-STANDBY
When Zulip is run behind one or more reverse proxies, you must
configure `loadbalancer.ips` so that Zulip respects the client IP
addresses found in the `X-Forwarded-For` header. This is not
immediately clear from the documentation, so this commit makes it more
clear and augments the existing examples to showcase this need.
Fixes: #19073
Alex Vandiver
Toyam Cox
David Rosa
Tran Sang
Tim Abbott
Anders Kaseorg
Emilio López