Matomo Load Balancing

Bluebird63 · July 1, 2020, 1:37pm

Hi All,

Hope everyone is safe and well. I’ve got a few quick questions for those in the know which might hopefully save me a bit of bother as i’m about to setup Matomo in a load balanced environment

The FAQ says to enable database storage. It looks like this has been enabled in Matomo by default since 3.8.0. However, i seem to have a lot of recent files in my \tmp\cache\tracker. Does this mean I don’t have this enabled? Is there a way I can check this?
I have the Query Tracking plugin up and running in an isolated server - is this compatible with Load balancer, i.e can a load balanced webservers use the same Redis DB without issue?

EDIT - a bit more on 1) I received this error when opening matomo on my new load balanced webserver; does this mean that my my temp sessions are not being stored at the db level?

The directory “XXXXXXXXXXXX/tmp/cache/tracker/” is not writable.

Thanks All, stay safe

Lukas · July 2, 2020, 10:19am

Hi,

You can check your config.ini.php for a line called session_save_handler. If it isn’t set then the database is used by default. (In Matomo 4 that option is gone)

github.com

matomo-org/matomo/blob/3.x-dev/config/global.ini.php#L403-L405


; Matomo uses PHP's dbtable for session.
; If you prefer configuring sessions through the php.ini directly, you may unset this value to an empty string
session_save_handler = dbtable

This only affects where sessions are stored. Matomo still stores other temporary things in the cache (e.g. information about marketplace plugins) which are stored in tmp/cache.

I can’t say much about load balancing as I have no experience with it.

Bluebird63 · July 2, 2020, 10:41am

makes sense, thanks for the quick reply Lukas

Bluebird63 · July 6, 2020, 8:15am

Hi @Lukas ,

I’ve found a bit more about this which you may be able to help with! @matthieu wrote this on the github page, to be added to the FAQ

github.com/matomo-org/plugin-QueuedTracking

QueuedTracking setup on multiple servers (new guide)

opened 05:02AM - 15 May 20 UTC

mattab

Here are some notes I wrote earlier and thought it would be useful to put in the… FAQ maybe? ## How do I setup QueuedTracking on multiple tracking servers? Say you have - 4 servers, - and setup 8 workers in QueuedTracking, - and disabled the option in the settings to process the queue during the tracking request, then on each of your 4 frontend servers, you need to: - setup 2 crontab to run each minute and which execute the command: `./console queuedtracking:process --queue-id=X` Where X is the queue ID. Each server handles 2 queues. So the 4 servers handle the 8 queues. Queue ID starts at 0. Notes: * execute the command `./console queuedtracking:monitor` to track the state of the queue * When using multiple workers it might be worth to lower the number of "Number of requests to process" to eg 15 in "Plugin Settings". By default 25 requests are inserted in one step by using transactions. This means different workers might have to wait for each other. By lowering that number each worker will block the DB for less time. * you can optionally receive an email when the number of requests queued in a single queue reaches a configured threshold. You can configure this in your config/config.ini.php config file using the following configuration: ```ini [QueuedTracking] notify_queue_threshold_emails[] = example@example.org notify_queue_threshold_single_queue = 250000 ```

It mentions create a specific queue (or 2) per webserver and setting up a crontab to run on each server to run the server’s assigned queue (makes sense, essentially assigning queues per webserver to insert into the db) with the following command;

./console queuedtracking:process --queue-id=X

When activating ./console queuedtracking:process --queue-id=0 specifically for queue-id=0, it doesn’t work, i get this error

ERROR [2020-07-06 09:10:58] 4700 Uncaught exception: C:\inetpub\wwwroot\vendor\symfony\console\Symfony\Component\Console\Input\ArgvInput.php(242): The “–queue-id” option requires a value.

It works for fine for “./console queuedtracking:process --queue-id=1”

is this a known issue or am i doing something incorrectly?

Many Thanks!
Steve

Lukas · July 6, 2020, 3:54pm

Hi,

For me both work (as I’d expect from the code)

➜  /var/www/matomo sudo -u www-data php7.3 /var/www/matomo/console queuedtracking:process --queue-id=0
Forcing queue ID: 0
Starting to process request sets, this can take a while

This worker finished queue processing with 0req/s (0 requests in 0.00 seconds)

➜  /var/www/matomo sudo -u www-data php7.3 /var/www/matomo/console queuedtracking:process --queue-id=1
Forcing queue ID: 1
Starting to process request sets, this can take a while

This worker finished queue processing with 0req/s (0 requests in 0.00 seconds)

Can you check if your shell is maybe modifying something and sending a different command (I also have to say I am not that familiar with Windows and Symphony Console that handles the command line arguments might be doing something weird)

Bluebird63 · July 6, 2020, 4:47pm

Hey @Lukas, thanks for getting back mate.

Not sure how I would go about checking that but, I’ve just tried the command in Powershell as well as Windows Command Line and i’m getting the exact same error so i’m thinking it’s not the shell. It only happens on –queue-id=0, it is fine for –queue-id=1 etc

Bluebird63 · July 6, 2020, 5:54pm

It might be something to do with this code in matomo;
Symfony could be equating the 0 passed through as boolean false, not an integer and throwing the RTE

    if (null === $value) {
        if ($option->isValueRequired()) {
            throw new \RuntimeException(sprintf('The "--%s" option requires a value.', $name));
        }

Although i’m pretty thick and not a PHP man at all so could be completely wrong

Bluebird63 · July 9, 2020, 1:21pm

Hi All,

If everyone else is stuck on this I’ve come around with a slightly cruder plan B

Basically i was worried that in a load balanced environment with 3 servers not talking to each other there may be inefficiencies when searching for the queue to process, maybe even enforcing a lock.

So basically rather than being able to assign a queue ID per server to process with ** queuedtracking:process --queue-id=xxx*, I’ve time banded the queuedtracking:process process to run every minute on different 20 second intervals per server. i.e 00:00:00 on server 1, 00:00:20 on server 2 and 00:00:40 on server 3.

Does the same job and shares around the burden - if anyone has any better ways, would love to hear it. I’m now up and running with matomo in a load balanced environment and feeling a large sense of relief - COVID spikes really pushed our incumbent setup to the limit!!