2024 Session II (c242)

Stations

• Nn has pad N09
• Apex joining for the first time
• Ef is out

Observing Notes

• Fringe test was successful in detecting fringes between Nn, Pv, On, Ys, Mh.
• Pv stowed at ~07:15 UTC until scan 576, stopping again at 20:20 UTC, back at scan 743, then stopped after an hour.
• Nn stowed due to wind at 16:20 UTC and until 23:40 UTC(c242a)
• Apex out initially due to power failure, on source ~13:30 UTC
• Ys lost scans 209-222
• Oct 12 Nn stopped at 14:07
• Pv on sky since 14:30 UTC (scan 320).
• c242c/d Ys very foggy, cloudy
• Pv stopped due to strong wind at 19:45 UTC.
• Oct 13 Nn started with scan at 05:45 but had acquisition problems 07:30-08:30

APEX Module

APEX data are on BHC%0141 in CF502. In order to hand-carry GMVA data from APEX to Bonn, data were consolidated from two Mark6 modules (2 x 8 disks) onto a more readily transportable set of 8 loose disks. Data of one polarization are in the standard per-disk subdirectory 'data', data of the other polarization are in 'GMVA_slot2'.

To mount the "two modules" contained on BHC%0141, use:

d281
# assuming that BHC%0141 is in slot 1:
fuseMk6 -r '/mnt/disks/1/*/data' /`hostname -s`_fuse/1
fuseMk6 -r '/mnt/disks/1/*/GMVA_slot2' /`hostname -s`_fuse/2

During unrelated tests at MPIfR, unfortunately the filesystem metadata of 1 out of the 8 disks got erased. During a later trip to APEX the missing 'GMVA_slot2' files of that disk were copied out from the still existing module there and are now integrated back into BHC%0141. The missing 'data' files of that disk were less trivial to recover.

# Make a low level backup of the wiped disk
root@mark6-08> cd /data/gmva2024_2/
root@mark6-08> dd bs=1M if=/dev/sdb of=apex-module-disk1-wiped.raw status=progress
root@mark6-08> chmod a-w apex-module-disk1-wiped.raw
root@mark6-08> fdisk -lu apex-module-disk1-wiped.raw
 #        Start          End    Size  Type            Name
 1         2048  15627857919    7.3T  Microsoft basic MPIH%024_5
 2  15627857920  15628052479     95M  Microsoft basic MPIH%024_5m

# Grab the XFS file system structure from an intact disk
root@mark6-08> cd /data/gmva2024_2/
root@mark6-08> losetup --read-only -o $((512*2048)) /dev/loop1 /dev/sdc
root@mark6-08> xfs_metadump -g -f -o -w -a /dev/loop1 apex-module-disk2-intact.xfs_metadump
root@mark6-08> losetup -D ; losetup -a
 
# Transplant XFS structure from intact disk onto wiped-disk raw content
root@fxmanager> cd /data/gmva2024_2/
root@fxmanager> dd bs=512 if=apex-module-disk1-wiped.raw \
   of=recovery-attempt.fs skip=2048 count=$((15627857919-2048+1)) status=progress conv=notrunc
root@fxmanager> dd status=progress conv=notrunc bs=512 count=1024 \
    seek=15627855872 if=/dev/zero of=recovery-attempt.fs  # appends a bit of 0x00 padding
root@fxmanager> losetup -v -o 0 /dev/loop0 recovery-attempt.fs
root@fxmanager> xfs_mdrestore -g apex-module-disk2-intact.xfs_metadump /dev/loop0
  2070 MB read
root@fxmanager> mkdir cloop ; mount /dev/loop0 ./cloop/ -txfs -oro
  # success!
oper@fxmanager> cd /data/gmva2024_2/ ; mkdir recovered_content
oper@fxmanager> cp -anv ./cloop/data/*.vdif ./recovered_content/
oper@fxmanager> mkdir recovered_GMVA_slot2
oper@fxmanager> cp -anv ./cloop/data/*.vdif ./recovered_GMVA_slot2/
   
# Restore content: init the half-wiped partitions, restore Mk6 metadata
root@mark6-08> mkfs.xfs -f /dev/sdb1
root@mark6-08> mkfs.xfs -f /dev/sdb2
root@mark6-08> mount /dev/sdb2 /tmp ; cp -av /mnt/disks/.meta/1/2/* /tmp; umount /tmp
# Add GMVA_slot2 data from new disk from post-GMVA APEX visit
# (though could also use ./recovered_GMVA_slot2/, but did not get to proceed
# with the low-level recovery attempts until after the post-GMVA APEX visit :P)
stop & start mk5daemon
root@mark6-08> cp -anv /mnt/disks/3/1/GMVA_slot2_copy/* /mnt/disks/1/2/GMVA_slot2/
# Add 'data' files from restored image
oper@mark6-08> cp -anv /data/gmva2024_2/recovered_content/*.vdif /mnt/disks/1/2/data/