[Mondo-devel] SHOULD NOT GET HERE!

classic Classic list List threaded Threaded
6 messages Options
Reply | Threaded
Open this post in threaded view
|

[Mondo-devel] SHOULD NOT GET HERE!

David C. Partridge
BG2: [Main] libmondo-stream.c->read_header_block_from_stream#1195: Bad
header block at 33152 K
DBG2: [Main] libmondo-verify.c->verify_afioballs_from_stream#732: YOU SHOULD
NOT GET HERE
DBG2: [Main] libmondo-verify.c->verify_afioballs_from_stream#733: Grabbing
the EXAT files
INFO: Wrong marker! (Should be BLK_START_AFIOBALLS, is actually
BLK_START_AFIOBALLS)
DBG2: [Main] libmondo-stream.c->read_header_block_from_stream#1195: Bad
header block at 33280 K
INFO: Wrong marker! (Should be BLK_START_AN_AFIO_OR_SLICE, is actually
BLK_START_AN_AFIO_OR_SLICE)

Processing has stopped at this point ...

What's the problem likely to be?

PS are any of the mondorescue folk watching this or am I talking to myself?

Thanks
Dave


------------------------------------------------------------------------------
Check out the vibrant tech community on one of the world's most
engaging tech sites, SlashDot.org! http://sdm.link/slashdot
_______________________________________________
Mondo-devel mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/mondo-devel
Reply | Threaded
Open this post in threaded view
|

Re: [Mondo-devel] SHOULD NOT GET HERE!

Bruno Cornec-4
Hello David,

David C. Partridge said on Wed, Dec 14, 2016 at 07:01:47PM +0000:
>BG2: [Main] libmondo-stream.c->read_header_block_from_stream#1195: Bad
>header block at 33152 K

Humm.

>DBG2: [Main] libmondo-verify.c->verify_afioballs_from_stream#732: YOU SHOULD
>NOT GET HERE

Never seen that.

>What's the problem likely to be?

Problem handling the tape of course, but not sure waht.

>PS are any of the mondorescue folk watching this or am I talking to myself?

I'm reading most of the e-mail but hadn't had time to work on MondoRescue since one month at least :-(
My job is taking most of my work and free time these days.
I'll start digging into unanswered ones asap, including yours.

However, if you try tape + iSCSI there is a high probability it fails as you're the first to do it !
I've had some success with SCSI attached tapes, including OBDR for which I added support years ago.
However, I think you try to do stuff that have not been experimented before (but I read very rapidlyyour previous mails sorry for that).

Best regards,
bruno.
--
Open Source Profession, WW Linux Community Lead     http://www.hpintelco.net
HPE EMEA EG FLOSS Technology Strategist http://www.hpe.com/engage/opensource
FLOSS projects:    http://mondorescue.org         http://project-builder.org
Musique ancienne?   http://www.musique-ancienne.org  http://www.medieval.org

------------------------------------------------------------------------------
Check out the vibrant tech community on one of the world's most
engaging tech sites, SlashDot.org! http://sdm.link/slashdot
_______________________________________________
Mondo-devel mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/mondo-devel
Reply | Threaded
Open this post in threaded view
|

Re: [Mondo-devel] SHOULD NOT GET HERE!

David C. Partridge
It's interesting that 33152 is exactly 384 (256 + 128) bytes beyond the end
of the largest block that the code could have used until yesterday,
 (and only then with my changes to mr_popup_and_get_string() which I'll
append here right now).

starting at line 768 of newt-specific.c:

        b_res = newtRunForm(myForm);
        //
        // if it worked ok allocate a new char array to hold the result
        // and copy the result to it
        //
        if (b_res != b_2) {
                out = malloc(1 + strlen(entry_value));
                strcpy(out, entry_value);
        }

        newtPopHelpLine();
        newtFormDestroy(myForm);
        newtPopWindow();
        mr_free(blurb);

        return(out);
}

The run you were looking at was using a tape block size of 65536 which
actually was large enough to get the tape streaming (32k is way too small).

Dave

-----Original Message-----
From: Bruno Cornec [mailto:[hidden email]]
Sent: 14 December 2016 22:27
To: Mondo mailing list
Subject: Re: [Mondo-devel] SHOULD NOT GET HERE!

Hello David,

David C. Partridge said on Wed, Dec 14, 2016 at 07:01:47PM +0000:
>BG2: [Main] libmondo-stream.c->read_header_block_from_stream#1195: Bad
>header block at 33152 K

Humm.

>DBG2: [Main] libmondo-verify.c->verify_afioballs_from_stream#732: YOU
>SHOULD NOT GET HERE

Never seen that.

>What's the problem likely to be?

Problem handling the tape of course, but not sure waht.

>PS are any of the mondorescue folk watching this or am I talking to myself?

I'm reading most of the e-mail but hadn't had time to work on MondoRescue
since one month at least :-( My job is taking most of my work and free time
these days.
I'll start digging into unanswered ones asap, including yours.

However, if you try tape + iSCSI there is a high probability it fails as
you're the first to do it !
I've had some success with SCSI attached tapes, including OBDR for which I
added support years ago.
However, I think you try to do stuff that have not been experimented before
(but I read very rapidlyyour previous mails sorry for that).

Best regards,
bruno.
--
Open Source Profession, WW Linux Community Lead     http://www.hpintelco.net
HPE EMEA EG FLOSS Technology Strategist http://www.hpe.com/engage/opensource
FLOSS projects:    http://mondorescue.org         http://project-builder.org
Musique ancienne?   http://www.musique-ancienne.org  http://www.medieval.org

----------------------------------------------------------------------------
--
Check out the vibrant tech community on one of the world's most engaging
tech sites, SlashDot.org! http://sdm.link/slashdot
_______________________________________________
Mondo-devel mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/mondo-devel


------------------------------------------------------------------------------
Check out the vibrant tech community on one of the world's most
engaging tech sites, SlashDot.org! http://sdm.link/slashdot
_______________________________________________
Mondo-devel mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/mondo-devel
Reply | Threaded
Open this post in threaded view
|

Re: [Mondo-devel] SHOULD NOT GET HERE!

David C. Partridge
Well, there went one theory, I re-did the test using a blksize of 32768
hoping that it would verify OK.

 No such luck.  :(  

It failed with the identical errors and the window appeared to be hung, but
a couple of Ctrl-Cs showed a message saying that mondoarchive failed with a
segfault :(

Please let me know what debug information I can gather.  I couldn't find the
core file :(  I think apport has eaten it (not sure how to stop that).

Dave

-----Original Message-----
From: David C. Partridge [mailto:[hidden email]]
Sent: 15 December 2016 02:39
To: 'Mondo mailing list'
Subject: Re: [Mondo-devel] SHOULD NOT GET HERE!

It's interesting that 33152 is exactly 384 (256 + 128) bytes beyond the end
of the largest block that the code could have used until yesterday,  (and
only then with my changes to mr_popup_and_get_string() which I'll append
here right now).

starting at line 768 of newt-specific.c:

        b_res = newtRunForm(myForm);
        //
        // if it worked ok allocate a new char array to hold the result
        // and copy the result to it
        //
        if (b_res != b_2) {
                out = malloc(1 + strlen(entry_value));
                strcpy(out, entry_value);
        }

        newtPopHelpLine();
        newtFormDestroy(myForm);
        newtPopWindow();
        mr_free(blurb);

        return(out);
}

The run you were looking at was using a tape block size of 65536 which
actually was large enough to get the tape streaming (32k is way too small).

Dave

-----Original Message-----
From: Bruno Cornec [mailto:[hidden email]]
Sent: 14 December 2016 22:27
To: Mondo mailing list
Subject: Re: [Mondo-devel] SHOULD NOT GET HERE!

Hello David,

David C. Partridge said on Wed, Dec 14, 2016 at 07:01:47PM +0000:
>BG2: [Main] libmondo-stream.c->read_header_block_from_stream#1195: Bad
>header block at 33152 K

Humm.

>DBG2: [Main] libmondo-verify.c->verify_afioballs_from_stream#732: YOU
>SHOULD NOT GET HERE

Never seen that.

>What's the problem likely to be?

Problem handling the tape of course, but not sure waht.

>PS are any of the mondorescue folk watching this or am I talking to myself?

I'm reading most of the e-mail but hadn't had time to work on MondoRescue
since one month at least :-( My job is taking most of my work and free time
these days.
I'll start digging into unanswered ones asap, including yours.

However, if you try tape + iSCSI there is a high probability it fails as
you're the first to do it !
I've had some success with SCSI attached tapes, including OBDR for which I
added support years ago.
However, I think you try to do stuff that have not been experimented before
(but I read very rapidlyyour previous mails sorry for that).

Best regards,
bruno.
--
Open Source Profession, WW Linux Community Lead     http://www.hpintelco.net
HPE EMEA EG FLOSS Technology Strategist http://www.hpe.com/engage/opensource
FLOSS projects:    http://mondorescue.org         http://project-builder.org
Musique ancienne?   http://www.musique-ancienne.org  http://www.medieval.org

----------------------------------------------------------------------------
--
Check out the vibrant tech community on one of the world's most engaging
tech sites, SlashDot.org! http://sdm.link/slashdot
_______________________________________________
Mondo-devel mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/mondo-devel


----------------------------------------------------------------------------
--
Check out the vibrant tech community on one of the world's most engaging
tech sites, SlashDot.org! http://sdm.link/slashdot
_______________________________________________
Mondo-devel mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/mondo-devel


------------------------------------------------------------------------------
Check out the vibrant tech community on one of the world's most
engaging tech sites, SlashDot.org! http://sdm.link/slashdot
_______________________________________________
Mondo-devel mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/mondo-devel
Reply | Threaded
Open this post in threaded view
|

Re: [Mondo-devel] SHOULD NOT GET HERE!

David C. Partridge
First off fix wrong_marker() to allow for the fact the marker_to_string
result is a static!

/**
 * Log (to screen) an erroneous marker, along with what it should have been.
 * @param should_be What we were expecting.
 * @param it_is What we got.
 */
void wrong_marker(int should_be, int it_is)
{
        char outstr1[MAX_STR_LEN] = "";

        strcpy(outstr1, marker_to_string(should_be));
        log_to_screen("Wrong marker! (Should be %s, is actually %s)",
outstr1, marker_to_string(it_is));
}

/* @} - end of streamGroup */

Now re-run the verify:

INFO: Verifying backups
INFO: running: mt -f /dev/st0 setblk 32768 >
/tmp/mondo.tmp.fkhcSw/mondo-run-prog-thing.tmp 2>
/tmp/mondo.tmp.fkhcSw/mondo-run-prog-thing.err
INFO: --------------------------------start of
output-----------------------------
INFO: --------------------------------end of
output------------------------------
INFO: ...ran just fine. :-)
DBG2: [Main] libmondo-stream.c->insist_on_this_tape_number#473: Insisting on
tape #1
INFO: Waiting while the tape drive rewinds
INFO: Waiting while the tape drive rewinds
DBG2: [Main] libmondo-stream.c->insist_on_this_tape_number#488: I assume
user has inserted it. They _say_ they have...
DBG2: [Main] libmondo-stream.c->insist_on_this_tape_number#492: OK, I've
finished insisting. On with the revelry.
DBG2: [Main] libmondo-stream.c->openin_tape#811: Opening IN tape
INFO: root is mounted at /dev/mapper/Charon--vg-root
INFO: That doesn't mean /dev/mapper/Charon--vg-root is the root partition.
It's just a debugging message. Relax. It's part of
am_I_in_disaster_recovery_mode().
DBG1: [Main] libmondo-devices.c->am_I_in_disaster_recovery_mode#167: Is this
a ramdisk? result = FALSE
INFO: running: dd if=/dev/zero bs=1024 count=16k 2> /dev/null | buffer -o
/dev/null -s 32768 -m 32m > /tmp/mondo.tmp.fkhcSw/mondo-run-prog-thing.tmp
2> /tmp/mondo.tmp.fkhcSw/mondo-run-prog-thing.err
INFO: --------------------------------start of
output-----------------------------
INFO: --------------------------------end of
output------------------------------
INFO: ...ran just fine. :-)
INFO:
DBG2: [Main] libmondo-fifo.c->open_device_via_buffer#117: Calling buffer ---
command = 'buffer -m 32m -p20 -B -s32768 -i /dev/st0 2>>
/var/log/mondoarchive.log'
DBG2: [Main] libmondo-fifo.c->open_device_via_buffer#124: Successfully
opened ('r') tape device /dev/st0
INFO: running: ps auxww | grep "buffer -m 32m -p20 -B -s32768 -i /dev/st0
2>> /var/log/mondoarchive.log" >
/tmp/mondo.tmp.fkhcSw/mondo-run-prog-thing.tmp 2>
/tmp/mondo.tmp.fkhcSw/mondo-run-prog-thing.err
INFO: --------------------------------start of
output-----------------------------
INFO: root     27273  0.0  0.0   4508   752 ?        S    09:14   0:00 sh -c
buffer -m 32m -p20 -B -s32768 -i /dev/st0 2>> /var/log/mondoarchive.log
INFO: root     27276  0.0  0.0   4508   752 ?        S    09:14   0:00 sh -c
ps auxww | grep "buffer -m 32m -p20 -B -s32768 -i /dev/st0 2>>
/var/log/mondoarchive.log" > /tmp/mondo.tmp.fkhcSw/mondo-run-prog-thing.tmp
2> /tmp/mondo.tmp.fkhcSw/mondo-run-prog-thing.err
INFO: root     27278  0.0  0.0  14228   932 ?        S    09:14   0:00 grep
buffer -m 32m -p20 -B -s32768 -i /dev/st0 2>> /var/log/mondoarchive.log
INFO: --------------------------------end of
output------------------------------
INFO: ...ran just fine. :-)
INFO: running: ps auxww | grep buffer | grep -v grep >
/tmp/mondo.tmp.fkhcSw/mondo-run-prog-thing.tmp 2>
/tmp/mondo.tmp.fkhcSw/mondo-run-prog-thing.err
INFO: --------------------------------start of
output-----------------------------
INFO: root     27273  0.0  0.0   4508   752 ?        S    09:14   0:00 sh -c
buffer -m 32m -p20 -B -s32768 -i /dev/st0 2>> /var/log/mondoarchive.log
INFO: root     27274  3.0  0.8  36976 34028 ?        D    09:14   0:00
buffer -m 32m -p20 -B -s32768 -i /dev/st0
INFO: root     27275  0.0  0.0  36976    80 ?        S    09:14   0:00
buffer -m 32m -p20 -B -s32768 -i /dev/st0
INFO: --------------------------------end of
output------------------------------
INFO: ...ran just fine. :-)
INFO: Buffer successfully started.
INFO: Reading stream
DBG2: [Main] libmondo-stream.c->openin_tape#819: stream device = '/dev/st0'
INFO: Skipping data disks on stream
INFO:     Skipping data disks on stream  
INFO: Skipping data disks on stream
INFO: Wrong marker! (Should be BLK_START_OF_TAPE, is actually
BLK_START_AN_AFIO_OR_SLICE)
INFO: Wrong marker! (Should be BLK_START_OF_BACKUP, is actually
BLK_START_FILE)
DBG2: [Main] libmondo-stream.c->openin_tape#866: Saved all.tar.gz to
'/tmp/mondo.tmp.fkhcSw/tmp/all.tar.gz'
INFO: Verifying regular archives on tape
DBG4: [Main] libmondo-tools.c->read_cfg_var#1069: Configuration item
last-filelist-number is 26
INFO: I am verifying archives against your live filesystem now.
INFO: Please wait. This may take a couple of hours.
INFO:
DBG2: [Main] libmondo-stream.c->read_header_block_from_stream#1195: Bad
header block at 33152 K
DBG2: [Main] libmondo-verify.c->verify_afioballs_from_stream#732: YOU SHOULD
NOT GET HERE
DBG2: [Main] libmondo-verify.c->verify_afioballs_from_stream#733: Grabbing
the EXAT files
INFO: Wrong marker! (Should be BLK_START_AFIOBALLS, is actually BLK_UNKNOWN
(-1))
DBG2: [Main] libmondo-stream.c->read_header_block_from_stream#1195: Bad
header block at 33280 K

Dave

-----Original Message-----
From: David C. Partridge [mailto:[hidden email]]
Sent: 15 December 2016 16:41
To: 'Mondo mailing list'
Subject: Re: [Mondo-devel] SHOULD NOT GET HERE!

Well, there went one theory, I re-did the test using a blksize of 32768
hoping that it would verify OK.

 No such luck.  :(  

It failed with the identical errors and the window appeared to be hung, but
a couple of Ctrl-Cs showed a message saying that mondoarchive failed with a
segfault :(

Please let me know what debug information I can gather.  I couldn't find the
core file :(  I think apport has eaten it (not sure how to stop that).

Dave

-----Original Message-----
From: David C. Partridge [mailto:[hidden email]]
Sent: 15 December 2016 02:39
To: 'Mondo mailing list'
Subject: Re: [Mondo-devel] SHOULD NOT GET HERE!

It's interesting that 33152 is exactly 384 (256 + 128) bytes beyond the end
of the largest block that the code could have used until yesterday,  (and
only then with my changes to mr_popup_and_get_string() which I'll append
here right now).

starting at line 768 of newt-specific.c:

        b_res = newtRunForm(myForm);
        //
        // if it worked ok allocate a new char array to hold the result
        // and copy the result to it
        //
        if (b_res != b_2) {
                out = malloc(1 + strlen(entry_value));
                strcpy(out, entry_value);
        }

        newtPopHelpLine();
        newtFormDestroy(myForm);
        newtPopWindow();
        mr_free(blurb);

        return(out);
}

The run you were looking at was using a tape block size of 65536 which
actually was large enough to get the tape streaming (32k is way too small).

Dave

-----Original Message-----
From: Bruno Cornec [mailto:[hidden email]]
Sent: 14 December 2016 22:27
To: Mondo mailing list
Subject: Re: [Mondo-devel] SHOULD NOT GET HERE!

Hello David,

David C. Partridge said on Wed, Dec 14, 2016 at 07:01:47PM +0000:
>BG2: [Main] libmondo-stream.c->read_header_block_from_stream#1195: Bad
>header block at 33152 K

Humm.

>DBG2: [Main] libmondo-verify.c->verify_afioballs_from_stream#732: YOU
>SHOULD NOT GET HERE

Never seen that.

>What's the problem likely to be?

Problem handling the tape of course, but not sure waht.

>PS are any of the mondorescue folk watching this or am I talking to myself?

I'm reading most of the e-mail but hadn't had time to work on MondoRescue
since one month at least :-( My job is taking most of my work and free time
these days.
I'll start digging into unanswered ones asap, including yours.

However, if you try tape + iSCSI there is a high probability it fails as
you're the first to do it !
I've had some success with SCSI attached tapes, including OBDR for which I
added support years ago.
However, I think you try to do stuff that have not been experimented before
(but I read very rapidlyyour previous mails sorry for that).

Best regards,
bruno.
--
Open Source Profession, WW Linux Community Lead     http://www.hpintelco.net
HPE EMEA EG FLOSS Technology Strategist http://www.hpe.com/engage/opensource
FLOSS projects:    http://mondorescue.org         http://project-builder.org
Musique ancienne?   http://www.musique-ancienne.org  http://www.medieval.org

----------------------------------------------------------------------------
--
Check out the vibrant tech community on one of the world's most engaging
tech sites, SlashDot.org! http://sdm.link/slashdot
_______________________________________________
Mondo-devel mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/mondo-devel


----------------------------------------------------------------------------
--
Check out the vibrant tech community on one of the world's most engaging
tech sites, SlashDot.org! http://sdm.link/slashdot
_______________________________________________
Mondo-devel mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/mondo-devel


----------------------------------------------------------------------------
--
Check out the vibrant tech community on one of the world's most engaging
tech sites, SlashDot.org! http://sdm.link/slashdot
_______________________________________________
Mondo-devel mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/mondo-devel


------------------------------------------------------------------------------
Check out the vibrant tech community on one of the world's most
engaging tech sites, SlashDot.org! http://sdm.link/slashdot
_______________________________________________
Mondo-devel mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/mondo-devel
Reply | Threaded
Open this post in threaded view
|

Re: [Mondo-devel] SHOULD NOT GET HERE!

Bruno Cornec-4
David C. Partridge said on Sat, Dec 17, 2016 at 09:33:59AM +0000:
>First off fix wrong_marker() to allow for the fact the marker_to_string
>result is a static!

You should use the latest source code version, which has wrong_marker using marker_to_string which does dynamic allocation of memory.

>INFO: Skipping data disks on stream
>INFO:     Skipping data disks on stream
>INFO: Skipping data disks on stream
>INFO: Wrong marker! (Should be BLK_START_OF_TAPE, is actually
>BLK_START_AN_AFIO_OR_SLICE)
>INFO: Wrong marker! (Should be BLK_START_OF_BACKUP, is actually
>BLK_START_FILE)
>DBG2: [Main] libmondo-stream.c->openin_tape#866: Saved all.tar.gz to
>'/tmp/mondo.tmp.fkhcSw/tmp/all.tar.gz'
>INFO: Verifying regular archives on tape

IIRC, there has been issues with Verify and tape for quite some time now, and I don't have acess to a tape drive atm to make tests with this and try to fix it.
But at the same time, the restore was working despite that ! I know it's weird, and I don't know what was done to break verify.

>Please let me know what debug information I can gather.  I couldn't find the
>core file :(  I think apport has eaten it (not sure how to stop that).

The best is to run valgrind + log files.
Now, in that case, this is probably an error with the logic of the program, rather than a code error tat valgrind can find :-(

Bruno.
--
Open Source Profession, WW Linux Community Lead     http://www.hpintelco.net
HPE EMEA EG FLOSS Technology Strategist http://www.hpe.com/engage/opensource
FLOSS projects:    http://mondorescue.org         http://project-builder.org
Musique ancienne?   http://www.musique-ancienne.org  http://www.medieval.org

------------------------------------------------------------------------------
Developer Access Program for Intel Xeon Phi Processors
Access to Intel Xeon Phi processor-based developer platforms.
With one year of Intel Parallel Studio XE.
Training and support from Colfax.
Order your platform today.http://sdm.link/intel
_______________________________________________
Mondo-devel mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/mondo-devel