Command-line access to F_SETLEASE
$ touch q
$ ./lease q rw -- sh -c 'echo Cleanup request && sleep 1 && echo Now clean'&
[1] 28330
$ echo "Before trying to access" && cat q && echo "After tying to access"
Before trying to access
Cleanup request
Now clean
[1] Done ./lease q rw -- sh -c 'echo Cleanup request && sleep 1 && echo Now clean'
After tying to access
Like flock
tool, can also operate on file descriptors.
Command-line access to control groups memory pressume monitoring events
Usage: cgroup_memory_pressure_monitor {low|medium|critical} /sys/fs/cgroup/memory/your_cgroup
Fill up memory with non-zero content compressible by Ultra Kernel Same page Merging.
Rundll for Linux. Loads main-like symbol from a library and executes it.
$ ./runso ./runso main ./runso main
Usage: runso ./libsomelibrary.so main <args>
Set oom_score_adj, then fill up memory with zeroes.
$ ./mempig
Killed
$ dmesg | grep -i oom
[228859.359178] mempig invoked oom-killer: gfp_mask=0x24280ca(GFP_HIGHUSER_MOVABLE|__GFP_ZERO), nodemask=0, order=0, oom_score_adj=1000
[228859.359234] [<ffffffff81168397>] oom_kill_process 0x237/0x450
[228859.359372] [ pid ] uid tgid total_vm rss nr_ptes nr_pmds swapents oom_score_adj name
[228860.221001] oom_reaper: reaped process 27664 (mempig), now anon-rss:0kB, file-rss:0kB, shmem-rss:0kB
Listed UDP port and do two things:
- Periodically send a UDP packet to specified destination (to keep NAT connection alive). The receiver is probably the udppair (see below).
- For each received UDP packet containing two numbers x and y (scanf "%d%d", &x,&y), write y to file descriptor number x.
Also tries to raise it's scheduling priority and lock all pages to memory.
Intended for remotely controlling various /sys/class/{brightness,gpio,pwm}/ knobs in a simplistic way.
Listen two UDP ports and exchange information between them. Each reply goes to the lastly received peer address. Something like:
socat udp-l:1234 udp-l:1235
But it does not "lock in" to one peer.
Measure UDP connection timeout in NAT by probing with different delays and testing if connectin still works or not. Server part replies to UDP packets based on timeout in incoming UDP packets. Client checks increasing timeouts until it detects missing reply packets, then "bisects" to find out more exact shut-off delay. Range is from 2 to 512 seconds. There may be running public server on vi-server.org:909 port.
$ udptimeoutchecker probe 0.0.0.0 0 vi-server.org 909
Trying with timeout 1 seconds...OK
Trying with timeout 2 seconds...OK
Trying with timeout 4 seconds...OK
Trying with timeout 8 seconds...OK
Trying with timeout 16 seconds...OK
Trying with timeout 32 seconds...OK
Trying with timeout 64 seconds...OK
Trying with timeout 128 seconds...FAIL
Trying with timeout 128 seconds...FAIL
Trying with timeout 128 seconds...FAIL
Trying with timeout 96 seconds...FAIL
Trying with timeout 80 seconds...FAIL
Trying with timeout 72 seconds...FAIL
Trying with timeout 68 seconds...FAIL
Trying with timeout 66 seconds...FAIL
Intermediate result: 64
Trying with timeout 1 seconds...OK
Trying with timeout 2 seconds...OK
Trying with timeout 4 seconds...OK
Trying with timeout 8 seconds...OK
Trying with timeout 16 seconds...OK
Trying with timeout 32 seconds...OK
Trying with timeout 64 seconds...OK
Trying with timeout 128 seconds...FAIL
Trying with timeout 128 seconds...FAIL
Trying with timeout 128 seconds...FAIL
Trying with timeout 96 seconds...FAIL
Trying with timeout 80 seconds...FAIL
Trying with timeout 72 seconds...FAIL
Trying with timeout 68 seconds...FAIL
Trying with timeout 66 seconds...FAIL
Intermediate result: 64
Trying with timeout 1 seconds...OK
Trying with timeout 2 seconds...OK
Trying with timeout 4 seconds...OK
Trying with timeout 8 seconds...OK
Trying with timeout 16 seconds...OK
Trying with timeout 32 seconds...OK
Trying with timeout 64 seconds...OK
Trying with timeout 128 seconds...FAIL
Trying with timeout 128 seconds...FAIL
Trying with timeout 128 seconds...FAIL
Trying with timeout 96 seconds...FAIL
Trying with timeout 80 seconds...FAIL
Trying with timeout 72 seconds...FAIL
Trying with timeout 68 seconds...FAIL
Trying with timeout 66 seconds...FAIL
Intermediate result: 64
64
Prepend each UDP packets with 12 bytes of nanosecond timestamp (8 bytes for seconds, 4 for nanoseconds) and 4 bytes of per-client sequence number and send them back.
$ ./udp_ts_seq_reply 0.0.0.0 1919&
$ socat -x -v - udp:127.0.0.1:1919 > /dev/null
123
> 2018/02/19 19:52:08.981246 length=4 from=0 to=3
31 32 33 0a 123.
--
< 2018/02/19 19:52:08.981712 length=20 from=0 to=19
00 00 00 00 00 57 2f 47 09 a8 bf 83 00 00 00 01 .....W/G........
31 32 33 0a 123.
--
ABC
> 2018/02/19 19:52:11.439685 length=4 from=4 to=7
41 42 43 0a ABC.
--
< 2018/02/19 19:52:11.440078 length=20 from=20 to=39
00 00 00 00 00 57 2f 49 24 fa c5 ed 00 00 00 02 .....W/I$.......
41 42 43 0a ABC.
--
DEF
> 2018/02/19 19:52:13.094035 length=4 from=8 to=11
44 45 46 0a DEF.
--
< 2018/02/19 19:52:13.094455 length=20 from=40 to=59
00 00 00 00 00 57 2f 4b 10 61 48 68 00 00 00 03 .....W/K.aHh....
44 45 46 0a DEF.
--
$ dump_tc_stats
Usage: dump_tc_stats <iface>
Dump interface statistics periodically. Don't forget to `tc qdisc add <iface> root ...` something
$ dump_tc_stats veth0
{"PACKETS":23514,"BYTES":2979825,"RATE_BPS":0,"RATE_PPS":0,"QLEN":0,"BACKLOG":0,"DROPS":0,"REQUEUES":0,"OVERLIMITS":0}
{"PACKETS":23526,"BYTES":2981342,"RATE_BPS":0,"RATE_PPS":0,"QLEN":0,"BACKLOG":0,"DROPS":0,"REQUEUES":0,"OVERLIMITS":0}
...