Supersedes ISE_v10_Checkpoint_merged_6.md and _5 (neither is overwritten). Everything in _5 still stands unless stated here, including its standing rules (section 1), the MediaProcessor and cleanISE_Clone notes (sections 4 and 5) and the carried ISE facts (section 7). Read _5 and this file together.
A Windows restart without shutting VMware down changed the IPs of Dev and Clone and corrupted their guest profiles. The clock on the guests was also far out. He then deleted Clone's VMware snapshot files, so Clone went back to the state of its base disk (older than before). A backup of those files exists on his 3 TB drive but is too old to be useful.
| Box | Old IP | New IP | Hostname |
|---|---|---|---|
| Live | unchanged | unchanged | forum.drugs-and-users.org |
| Dev | 192.168.233.129 | 192.168.233.142 | dev.mirror |
| Clone | 192.168.233.140 | 192.168.233.135 | clone |
| File | Version | md5 | Notes |
|---|---|---|---|
ise_settings_v4.2.zip |
4.2 (was 4.1) | de379f79375dc86ba2174625a9a5ed1e |
Only the six Dev/Clone rows of _SITE_ROOTS changed (Dev and Clone, each main/smf20/smf21). Live rows untouched. Tested in a throwaway tree. |
ISE_indexers_v1.0.0.zip |
MiscIndexer 7.4.1#6, PDFIndexer 8.7.1, PDF-32bit-Indexer 8.6.1 | 1f9c8262ae22640df4e7c2a89eb0b4a9 |
The hard-coded DEV_IP line is gone. IS_DEV is now socket.gethostname().lower().split(".")[0] == "dev". IS_DEV only chooses the log file name (*_dev.log). get_machine_ip() is still defined and now unused. |
IndexBuilder.py v7.4.5 has no IP and is unchanged.hostname printing dev.mirror or dev. Not confirmed. Claude chose the hostname route without his go-ahead (disclosed to him)./bin scripts: json_md5sums_all (lines 18 and 20), copies_Dev_to_Clone, copies_Clone_to_Dev, scp.text. Clone is the master of all three /bin directories, so Claude needs these four files from Clone's /bin to patch them.ISE_results.html: stale output, and an http:// vs https:// mismatch whose source is not found. Next check: grep -n boardurl /var/www/html/Settings.php on Dev and Clone, or ResultFormatter.py.$boardurl in Dev and Clone Settings.php: change it through the SMF admin panel, never by editing the file.ISE_Data JSON stores: not searched.py and python3 were aliases to python3.15. Cron and #!/usr/bin/env python3 scripts use the real PATH: Live /usr/local/bin/python3 -> python3.8 (3.8.18). The indexers therefore still run under 3.8, which has pymysql; 3.15 does not (py MediaIndexer.py failed, py38 MediaIndexer.py worked: 2896 items, one added, instagram-1792390459852.mp4)./usr/local/bin/python3.15 running /opt/glibc-2.17/lib/ld-2.17.so with library path /opt/glibc-2.17/lib:/lib64:/usr/lib64:/usr/local/lib:/usr/local/openssl-1.1.1/lib, which runs /opt/python-3.15.0/bin/python3.15.python-3.15.0, openssl-1.1.1w, python3.8, Python-3.8.18 and Python-3.8.18.tgz. There is no 3.15 tarball, so 3.15 cannot be rebuilt from source there..real and .orig) in /usr/local, working. Clone had 3.15 before the rollback (he installed it there first); it is now gone (python3.15: command not found). Clone's /opt also lacks openssl-1.1.1w, python3.8, Python-3.8.18 and the tarball.Python-3.15.0.tgz, 43,686,280 bytes) in ~/Python-3.15.0, using the recipe in the Chapter 2 PDF: source /opt/rh/devtoolset-9/enable BEFORE configure (otherwise gcc 4.4 is used and -std=c11 fails); /root/build_py/shim_memalign.h (guarded with #ifndef __ASSEMBLER__, declares memalign); ./configure --prefix=/opt/python-3.15.0 --with-openssl=/usr/local/openssl-1.1.1 --with-openssl-rpath=auto CPPFLAGS="-include /root/build_py/shim_memalign.h -Daligned_alloc=memalign"; make -j3; make altinstall; wrapper /usr/local/bin/python3.15 as on Live. Six optional modules missing (_dbm _decimal _sqlite3 _tkinter _uuid _zstd), 0 failed. Tests passed (ssl 1.1.1w, blake2, pip 26.2.1)./opt/glibc-2.17/etc/localtime, /opt/glibc-2.17/share/zoneinfo); python3.15 now reports AEST/AEDT. They also change python3.8 timestamps on Clone to Sydney time.py alias should be made to match reality once 3.15 is back.vg_clone with PVs sda2, sda3, sda4; lv_root, lv_swap. MBR allows four primary partitions only, so sda4 was grown in place (fdisk), then pvresize and lvextend -r. Root went from 98% to 67% (95 GB total, about 30 GB free)./root/sda.table.orig (restore with sfdisk /dev/sda < /root/sda.table.orig)./etc/fstab on Clone is stale (LABEL=DOROOT /, /swapfile swap, no /boot); /rsync-live-root is a copy of Live's root on Clone. /boot unmounted and fstab tidy: do later, on a VM copy./root/.command_log timestamps are NOT reliable (clock was wrong; entries dated 18 Oct 23:38 and 20 Oct 05:55 came from today). Real facts: no 3.15 build commands from 18 Oct are in the log; /bin/mirrorISE_Clone is dated 16 Oct 22:48; /bin/cleanISE_Clone is not on Clone (never confirmed installed). Next: grep -m3 -i version in MediaProcessor.py (expect 9.30) and MediaIndexer.py (expect 9.9.9), then comm listings against Live. He has the delivery zips and will reinstall; timing is his call. Hold off cleanISE_Clone --apply and mtime-based mirrors until after about 05:55 on 20 Oct: files written while the clock was ahead carry future dates._5 deliveries survived on Clone: cleanISE_Clone v1.0.4, MediaProcessor v9.30 (checkpoint renamed to .ckpt), and the cron line. Their deploy status was never confirmed even before the rollback. Rebuild from the zips in _5 section 3 if gone./bin scripts (needs the four files) and find the source of the stale http:// links (section 3).hostname output for the new IS_DEV logic.swapon -s; ls -lh /swapfile._5: other readers of the old checkpoint name (grep -rIl processor_checkpoint /var/www/html /bin), AudioProcessor not yet seen, and the parked items._5 section 7: Clone's system glibc is 2.12; glibc 2.17 lives in /opt/glibc-2.17 and is used through the launchers.