-
Notifications
You must be signed in to change notification settings - Fork 86
remote_ssh : ERROR: src/utils/file.c:529: proceeds 0 bytes instead of 8 #44
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Comments
Hello!
probably, but I`m not sure |
Oh, I`ve forgot, there is now --remote option, which must be added to cmdline. |
You are also incorrectly using --remote-path options. |
Hello, thanks, it would have to be just : --remote-path=/var/lib/postgresql . I'll test again. |
I don`t think this is a right way. Remote binary can have a different name, so --remote-path should point to binary, not directory. |
Ok, I am testing against are large testdb :
./pg_probackup/pg_probackup init -B /var/lib/pgbackup/pgprobackups |
hello, postgres@smadb2cs:~$ ./pg_probackup/pg_probackup -V |
fixed |
--remote option is now obsolete |
./pg_probackup/pg_probackup add-instance -B /var/lib/pgbackup/pgprobackups -D /var/lib/pgsql/data --remote-proto=ssh --remote-host=10.9.0.77 --remote-path=/var/lib/pgsql/pg_probackup/. --instance testdb this worked. also backup with --stream seems to start. pls tell how's the archive command supposed to look like for remote? I cannot figure it out. it always asks for local -B. should I create an instance in the PostgreSQL host as well? |
tried backups like : ./pg_probackup/pg_probackup backup -B /var/lib/pgbackup/pgprobackups/ --instance testdb -b FULL --stream ^^^ this seems to work. Just interrupted it. postgres@smadb2cs: ^^^ this I cannot figure out how to setup archive_command for remote currently testing : trying with : this will take hours.. but pls tell me about archive-push from the pgsql host to the catalog host. |
still running , I dont see -j 8 to be respected, current speed below what this line can do : 8.2MB/s. |
delete does not delete back up, seems to work but info still shows backup. |
yes, we will fix that |
What capacity this line has? |
Can you also please monitor CPU usage at your system (both at database and backup nodes)? |
I believe this can go up to 15MB/s , at least I think I got those numbers with pgbackrest, although this line is pretty messed up, I don't trust what the provider says . I want to have this FULL backup finished so to be able to test PAGE and DELTA. No special encryption , just : BTW any news on the archive-push from the remote PgSql server? |
CPU usage is low, less than 10% on 4 core VM |
What about disk usage on database and backup node? |
also there is a typo : this backup_id should be -i backup-id |
as expected, full backup should consume about 1.3 T , will let you now about that. for the time being the show stopper is the archive_command from the remote and how to configure : archive-push (params to use). thanks again. |
Is it possible to ask you to attach debugger to running instance of pg_probackup (at database host if you are executing "backup" command) and run the following command: thread apply all bt |
I mean I/O utilization on database and backup node, can you lookup that? |
just checked with iotop on the database host , it is about 8M/sec |
`(gdb) thread apply all bt Thread 2 (Thread 0x7f9c47e4d700 (LWP 2204)): Thread 1 (Thread 0x7f9c4adbe700 (LWP 2202)): |
For the time being -j is not such a major issue. However, archive_command is a must. Also I got huge rush with other tasks, talk to you later ppl! Thanks! |
archive-push issue will be resolved soon. |
Sorry, but looks like you have attached debugger to pg_probackup instance at backup host (host from which you have launched remote backup). Can you please attach debugger to this process and get stacktraces of its threads? |
`(gdb) thread apply all bt Thread 3 (Thread 0x7f930adb9700 (LWP 127313)): Thread 2 (Thread 0x7f930a5b8700 (LWP 127314)): Thread 1 (Thread 0x7f930dd2a700 (LWP 127308)): |
Did you specify -j 8 options? |
in this run, I did not specify -j. But when I had run with -j 8 I saw no difference in number of procs running. |
Number of of processes will be the same, but there should be several backup_files threads in one process (unlike Postgres itself, pg_propackup is multithreaded). |
wow! thanks! still I see 3 threads although I did not specify -j . |
Except main thread, there is one thread for streaming WAL and one or more threads for copying data files. It will be nice if you can measure speed with larger numbers of threads and produce pg_probackup profile or at least stack traces of all threads. |
Thank you, I will let this first full backup run, and then try some DELTA/PAGE with -j 4 . |
Hello, As I told you this is a very unreliable line. This is supposed to be between two Swiss cloud providers in Switzerland, but still it sucks. So, we can't live without resume. Thank you for all your effort you put in this, and I hope to come back to use pg_probackup! |
Well, resume is not hard to implement, so I think it will be in 2.0.27 |
Thanks. |
Hello, I hope I get some news on that, and also I hope you'll enjoy the article. |
Hello! |
Hello
I got a little confused, so I'll use :
remote_ssh branch for the backup on the backup server and
remote_pull branch for archive_command on the postgresql server
?
…On Fri, 8 Feb 2019 at 08:14, gsmol ***@***.***> wrote:
Hello!
Please, try remote_pull branch.
For a archive-push setup just add --remote-proto=ssh --remote-host=host
--ssh-options="-l remote_username" to archive-push cmdline.
Resume is still in WIP stage.
—
You are receiving this because you authored the thread.
Reply to this email directly, view it on GitHub
<#44 (comment)>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/APDXikF0SLKzhZr7S7BsFE5EFIPd1SaJks5vLRWogaJpZM4aLTYr>
.
|
remote_ssh is obsolete, use only remote_pull |
okie!
…On Fri, 8 Feb 2019 at 09:03, gsmol ***@***.***> wrote:
remote_ssh is obsolete, use only remote_pull
—
You are receiving this because you authored the thread.
Reply to this email directly, view it on GitHub
<#44 (comment)>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/APDXirEinZ8DN7LAFHaQeVh3bUK1OWJ8ks5vLSE2gaJpZM4aLTYr>
.
|
ok checkout remote_pull, worked postgres@smadb2cs:~$ cat /var/lib/pgbackup/pgprobackups/backups/testdb/pg_probackup.conf Backup instance informationpgdata = /var/lib/pgsql/data Remote access parametersremote-proto = ssh but ./pg_probackup/pg_probackup backup -B /var/lib/pgbackup/pgprobackups/ -b FULL --instance testdb --stream ??? on the pgsql logs I see no access to anything .. |
You need to provide parameters for libpq connection: |
Hi
those were not needed when I had tried with remote_ssh branch.
Those :
./pg_probackup/pg_probackup add-instance -B /var/lib/pgbackup/pgprobackups
-D /var/lib/pgsql/data --remote-proto=ssh --remote-host=xx.xx.xx.xx
--remote-path=/var/lib/pgsql/pg_probackup/. --instance testdb
and
./pg_probackup/pg_probackup backup -B /var/lib/pgbackup/pgprobackups/
--instance testdb -b FULL --stream
and
./pg_probackup/pg_probackup backup -B /var/lib/pgbackup/pgprobackups/
--instance testdb -b FULL
used to wotk. Since we provide local access to the -D locally on the pgsql
cluster site, why do we additionally need to specify again postgresql
access? And if we need to do so, why not specify -h localhost which would
mean from the remote xx.xx.xx.xx ?
…On Sat, 9 Feb 2019 at 10:51, gsmol ***@***.***> wrote:
You need to provide parameters for libpq connection:
Connection options: -U, --username=USERNAME user name to connect as
(default: current local user) -d, --dbname=DBNAME database to connect
(default: username) -h, --host=HOSTNAME database server host or socket
directory(default: 'local socket') -p, --port=PORT database server port
(default: 5432) -w, --no-password never prompt for password -W, --password
force password prompt
—
You are receiving this because you authored the thread.
Reply to this email directly, view it on GitHub
<#44 (comment)>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/APDXikPNl0gKptiijbaANVHq8NGY8YOIks5vLowBgaJpZM4aLTYr>
.
|
tried with all the libpq params (passwd via .pgpass, verified that it works) : ./pg_probackup/pg_probackup backup -B /var/lib/pgbackup/pgprobackups/ --instance testdb -b FULL -h xx.xx.xx.xx -d postgres -U postgres -p 5432 same with --stream |
bump, any news on this? any way to do remote backups? this used to semi-work in the old remote_ssh branch. |
It is in development and very unstable |
ok, if its out I'd be delightful to test. In the meantime I included
pg_probackup in the blog article I wrote. When its out I'll send it to you
guys to comment!
…On Sat, 16 Feb 2019 at 14:39, gsmol ***@***.***> wrote:
It is in development and very unstable
—
You are receiving this because you authored the thread.
Reply to this email directly, view it on GitHub
<#44 (comment)>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/APDXiltI3K68C5dlVmowkgtXnXSzRVNVks5vN_wNgaJpZM4aLTYr>
.
|
We appreciate you efforts, thank you! |
I think you will find it interesting that remote backup via ssh was merged in master and will be released on this week. |
great!
…On Tue, 16 Apr 2019 at 13:53, gsmol ***@***.***> wrote:
I think you will find it interesting that remote backup via ssh was merged
in master and will be released on this week.
—
You are receiving this because you authored the thread.
Reply to this email directly, view it on GitHub
<#44 (comment)>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/APDXilkgPtMjUj1tFlB6XeWlt6LLd_3fks5vhavFgaJpZM4aLTYr>
.
|
Are remote capabilities enabled now? This are the rpms i have pg_probackup-10-2.0.26-1.d8553c06afff82a3.x86_64 |
Yes, they are available in 2.1.1.
|
Problem has been solved by the packages update. Too old task. |
Hello,
just started to test remote capabilities and I get :
/pg_probackup/pg_probackup add-instance -B /usr/local/var/lib/pgsql/mypgprobacks --remote-proto=ssh --remote-host=192.168.1.109 --remote-port=5432 --remote-path=/var/lib/postgresql/pg_probackup -D /var/lib/postgresql/datatest --instance testdblin
ERROR: src/utils/file.c:529: proceeds 0 bytes instead of 8
the catalog host is FreeBSD and the remote is linux. Could that pose a problem?
The text was updated successfully, but these errors were encountered: