On Sunday 20 of December 2020 21:28:11 Michael via tde-users wrote:
On Sunday 20 December 2020 01:34:29 pm Slávek Banko
via tde-users wrote:
> Dne Sunday 20 of December 2020 18:12:25 Michael via tde-users
napsal(a):
On Sunday 20 December 2020 12:02:26 am Michele Calgaro
via tde-users
wrote:
On
2020/12/20 04:00 AM, Slávek Banko via tde-users wrote:
> We may consider placing the primary archive in a hosting center
> - for example, by purchasing larger disks for tde-box or by
> increasing the space provided for tde-vps. However, it will
> probably be a good idea to consider such changes only after we
> are able to manage the project funds and accept donations.
This is defintely something we will need to do in future, to move
away from possible one-person-dependant services. I agree with you
we need to figure out how to handle donations and project funds,
then we can relook at this.
In that vein (away from possible one-person-dependant services):
- How big is the archive? (<250GB? ~4TB?)
- How simple is it to back it up? (a few rsync’s? tarball?)
I have no issues storing a copy, or two, of the archive on one of my
backup servers (which are in hosting centers).
I'm glad you're interested in the details.
Current size and occupied space of the partition:
/dev/mapper/lvm1-tde_data 400G 336G 65G 84% /srv/tde
A subset of this volume is synchronized to the mirror system. First to
the primary mirror, from which the other mirrors are then
synchronized. This synchronization usually takes place several times a
day, with pauses of 3 hours between the individual synchronizations.
The size always decreases significantly at the moment of release, when
the contents of the official repository and PSB are identical. The
size of the repository is now constantly growing again as the packages
in PSB are built for R14.0.10~pre. Such an increase can be almost 100
GiB. At the same time, there is a gradual increase in volume when
adding support for new distributions.
During the outage, which I announced in early December, I replaced my
home server's hardware with slightly newer components, including the
purchase of new, larger disks. So now I have a sufficient reserve of
disk capacity again.
The specified partition is on two disks (RAID1). In addition to these
disks, I make a regular daily local backup on an external usb disk.
The backup disk requires significantly more capacity, as the daily
differences can be large - from hundreds of MiBs to tens of GiBs. Next
to this, I'm making a copy of the archive on my next machine in a
remote location (two disks in RAID1). Copying to this remote location
is not regular yet.
Hi Slávek,
Lol, interesting, but TMI ;)
- What’s the max size you’ve ever needed?
The current partition size is to be understood as "beginning size". Each
newly added distribution adds 1.5 to 12 GiB (depending on the number of
architectures). Once we release R14.1.x, it will mean new repositories, an
estimated new 80 GiB. So here it is necessary to take into account that
the required space will continue to increase.
- How many unique backups do you want? *
I use dirvish for backup. It uses the principle of hardlinks so that each
backup serves as an independent full backup, although it is performed as
an incremental one. Due to this independence, backups can have different
expiration times. A normal daily backup has an expiration date of 15 days,
weekly backups have an expiration date of 3 months, and the first weekly
backup of the month has an expiration date of one year.
- Rsync fine? Push? Pull?(what are the paths and
creds?) **
- cron every 4 hours okay?
An rsync server is provided for synchronization to the primary mirror. So
the primary mirror does the pull. As I mentioned, there are some parts
excluded from syncing to mirrors.
If we want to solve the backup of the entire primary archive and at the
same time keep the exclusion so that the mirror system does not grow
beyond the accepted limit and contains currently useful data, it is
possible to set the second path for rsync.
It’ll take half an hour to setup, and I guess we can
stop spamming the
rest of the list, so just send to me personally for the rest...
It will certainly be good for Mike, as the administrator of the primary
mirror, to take part in the discussion here as well.
Best,
Michael
michael@
* I’ll do one here at home off of one of my servers as well.
** I prefer pull, but just from a security issue of not giving people
access to my servers.
PS: I’ll send you a copy of my final script so you can automate the
remote location you were talking about too.
____________________________________________________
Cheers
--
Slávek