Server in Germany up & running? Replication packets?

replication
musicbrainz-vm
Tags: #<Tag:0x00007f2a1113d200> #<Tag:0x00007f2a1113d0c0>

#1

It looks like your server in Germany is up & running?

I can’t get any newer replication packets for your official Virtual Machine.
Replication packet #99848 not available
Do we just don’t have any newer packets generated for now?


#2

Replication is generally working, but the “historical” replication packets from the old servers aren’t available at the moment. 99847 was the last regular replication packet created at the old host; then another packet, 99848, was created after the site went read-only. It could only be downloaded for a short while afterwards, before the MetaBrainz site was taken offline.

The packets from 99849 onwards are available, but aren’t any use if you are missing 99848. You will therefore need to wait until the older packets will have been restored, probably this (European/UTC) evening.


#3

OK, I see. Thank you for the explanation @chirlu
(I missed the packet #99848).

I will wait until tomorrow.


#4

Unfortunately, it’s more complicated than previously thought and packet 99848 may have been lost completely. In that case, the only way forward would be to make a fresh import of today’s dump, and continue replication from that point.

If you urgently need the current data, that (dropping DB, fresh import) is the best you can do. If you are not in a hurry, you can wait and see if we can find or recreate 99848, but there are no guarantees.


#5

Ups…
Thank you for the update @chirlu

If I drop my actual DB, download a complete dump (from where?) and make a fresh import of the data, do I have to recreate all my search indeces too?

After this steps, the replication should start at which packet number?


#6

You can get the dump from the usual FTP server. You’ll need today’s dump (20161111-041941), or any later one when more will have been created. Today’s dump should include everything up to replication packet 99872, so the replication would continue with 99873. As for search indexes, that’s orthogonal to importing a new dump – the old indexes would continue to work, but are of course outdated; but that’s no difference to working replication, the search indexes will always stay behind.

Do you have the optional cover_art_archive schema in your replicated DB? If so, you could help us by dumping its contents, in the state it is in now (i.e. after packet 99847), and putting it up somewhere for us to download. (The dump file is probably large, but should benefit heavily from compression using zip, gzip, xz or similar.)

The command would be something like pg_dump >caa-dump.sql -n cover_art_archive musicbrainz_db (see pg_dump documentation for more background).


#7

[quote=“chirlu, post:6, topic:170840”]Do you have the optional cover_art_archive schema in your replicated DB?[/quote]Sorry, seems to be empty :fearful:


#8

Too bad. :frowning: Thanks for trying, though.


#9

Latest news: It seems packet 99848 has been found. It will hopefully be available in the regular way later today or tomorrow.


#10

Any news about this crucial packet# 99848?


#11

Not yet. But we got all the others, and we also got the GPG signature for 99848. :wink: Not that that would be too useful …

We’ll have to see.


#12

I, too, am waiting for 99848. :slight_smile:

But I do have the cover_art_archive schema:
“name”,“row_count”
“cover_art_type”,1160392
"index_listing",1082896
"release_group_cover_art",3321
"cover_art",1082896
"art_type",13
"image_type",4

I can provide the dump file if you still need it.


#13

Thank you – it would be great if you could keep the data for now (i.e. not drop the database for a reimport), and I will contact you if we need a dump. There is still hope that we can get back the original packet, which of course would be the best solution (whereas trying to recreate the replication packet is a complicated endeavour).


#14

Has there been any update to this lost/missing packet?

It feels odd having my Musicbrainz stuck on Nov 8.


#15

Doesn’t really contain anything new beyond what I said before.


#16

Truly heroic efforts. Thanks for all of your efforts.


#17

We now have a process to recreate replication packet 99848, as soon as we have a database stuck at 99847. So I’ve written @mbcz an email and hope that part will soon be solved.


#18

I sent you a link via email to the full dump.


#19

@chirlu I got a cover_art_archive dump from a friend:


I have no idea, if this copy is useful for you…


#20

I have a database with 99847 as the last replicated packet but my upload speed would be slow.
If you want access to it let me know.