MetaCPAN Undergoes "Unplanned Re-index"

Earlier today a some dev work on MetaCPAN had the unfortunate side effect of dropping the ElasticSearch index. This has effectively forced us to re-index the site from scratch. What makes this particularly painful is that we've lost the author data (including +1) which so many Perl devs have lovingly created over the past few weeks.

The short story is that minicpan should be indexed within a few hours. The rest of CPAN and BackPAN will likely be in the next day or so.

Because we've been looking at adding new features and haven't spent time on a proper backup strategy, we're in a suboptimal place right now. The good news is that this didn't happen several weeks from now with much more author data. Also, we're now forced to find a workable backup solution as our first priority.

I think now would be a good time to put out a call for volunteers. If you're inclined to help dig in with some sysadmin-minded tasks or just want to help with the overall MetaCPAN ecosystem, we could really use your help. The interest in the project is great, but so is the number of open issues, which grows by the day.

The more devs we have who can dig in and get something done, the closer we'll be to getting a bullet-proof system online which we can deploy and recommend with confidence.

Now might be a good time to remind folks that we are indeed still in beta, so there is some allowance for unplanned downtime etc. We're doing our best with the resources we have. If you'd like to help speed the progress, your contributions will be warmly welcomed.

Please drop in at #metacpan on irc.perl.org with kudos or complaints.

9 Comments

Is there a list of issues somewhere sorted or grouped by priority (maybe the same as github issues, maybe not)?

Appreciate the heads up, I'll do what I can to help out.

Can you provide more details about what went wrong? I think data-loss on that scale is too large an issue to explain away with a vague "unfortunate side-effect".

The time you spent building & promoting it, plus the time people spent eagerly filling out their profiles needs a more detailed explanation to restore confidence, I think.

Hi Olaf

"...we've been looking at adding new features and haven't spent time on a proper backup strategy...".

This is a classic statement of project management failure :-).

Top 3 priorities are - always -:
o Reliability.
o Reliability.
o Reliability.

A non-existent or dysfunctional service is nowhere as good as anybody thinks it is.

That's my contribution....

Cheers
Ron

Well, I've personally been very little help since kicking in my $20 over beers at Burgundy's -- so count me in to help out, even if it's just contributing $s.

For those of you who are complaining, I say: put your money where your mouth is.

Where can we contribute?

I got quite upset about this when I first read but I can understand that stuff might happen. Especially to people who do things.

With that said I think it is very important to make sure people feel confident that data they enter will not be lost again. Neither by accident nor due to planned upgrade.

The reason for this is that you want people to engage with the system either by coding or at least by clicking on +1s. I am now a lot less enthusiastic about adding my +1s again as I don't like to do the same thing twice. Eventually I'll do this but it would be much easier if I had an assurance or at least a clear plan that I won't need to do this again.

Hey, someone could probably write a script using the MetaCPAN::API so each one of us could fetch and backup their own personal data (including their +1s) and then submit them if the database gets dropped again.

Not really, Ron.

You will simply fail to get off the ground if you insist on dotting all your Is and crossing all your Ts too early. I have the tendency to do too much of that and it’s not helpful, believe me.

It won’t work if you work shoddily, of course, but with finite resources you have to take calculated risks and accept growing pains to move fast and grow fast. If you take the right risks, you will gain the resources to take care of omissions soon; play it safe and it will take a very long time to get anywhere.

That’s the same tightrope walk that makes startup founders who do it well rich in a short amount of time… under harrowing conditions. The MetaCPAN folks are accidentally getting a training version of the same experience…

Leave a comment

About Olaf Alders

user-pic I hack on iCPAN, MetaCPAN and other fun stuff.