ead6cb918a
log backfill: synthesize timestamps when in 1970s
...
if the log message timestamp is in 1970, treat it as if
the epoch were the system boot time
2025-10-28 23:29:20 +00:00
f4b87430ea
tai64: write timestamp
2025-10-28 23:28:33 +00:00
c3d6ad7ad4
unix time to TAI conversion
2025-10-28 21:05:55 +00:00
8049714cdb
rename anoia.tai64.to-utc as to-unix
...
it's a far more accurate name for somethingthat return non-leap
seconds since 1-1-1970
2025-10-28 20:17:35 +00:00
8106389b74
add log backfill services
2025-10-27 21:19:43 +00:00
0f90f3f46d
lualinux.open needs three args
2025-10-27 21:18:52 +00:00
481715f906
remove debug output
2025-10-27 21:16:49 +00:00
c29416a287
move log fifo away from /run/log which doesn't exist yet
2025-10-27 21:16:17 +00:00
9e042bd385
ppp: ignore error return from writing readiness notification
...
ip-up and ipv6-up both attempt to write to fd 10, but only the
first writer will succeed because the peer closes the pipe as soon
as it gets anything. I can't see a way to make this race-proof, so
... here we are
2025-10-27 21:12:06 +00:00
26e1608fa7
align log shipping start/stop messages between producer and consumer
...
also there's no need to make it configurable, so get rid of that
2025-10-26 20:41:19 +00:00
ad9bba4f93
logtap: add WIP backfill command
2025-10-26 20:34:52 +00:00
6f559e3700
add fennelc, a script to do what writeFennel does
...
it's a bit simpler because no need to do quite as much string surgery
to compute package.{c,}path
also it can be used in Makefiles
2025-10-26 20:32:52 +00:00
6478c56ed1
log.shipping.command is now a command not a string
2025-10-25 17:34:50 +01:00
186543d2de
remove option for log shipping socket
...
* it's a fifo now not a socket
* I think it's an internal detail, not public interface
* backfill will need a second fifo anyway
2025-10-24 23:53:46 +01:00
c8b3188bd4
remove unneeded module
2025-10-24 23:46:13 +01:00
60ff3ab688
bordervm espose 443 for logs not 9428
...
this is because nginx is proxying them.
probably we should pick a non-default port all the same, but need to
investigate what's convention here
2025-10-20 23:48:21 +01:00
bfb4799075
certifix-client provice ca-certificate as an output
2025-10-20 23:47:21 +01:00
1b14161a0e
modern openssl needs rsa key >= 2048
...
per
https://docs.openssl.org/3.3/man3/SSL_CTX_set_security_level/#default-callback-behaviour
the default "security level" in openssl 3.3 is 2
2025-10-20 23:45:53 +01:00
72acb2e711
Merge remote-tracking branch 'newedge/fix-gl-ar750'
2025-10-19 14:22:49 +01:00
87c45dec67
switch certifix-client to luahttp
2025-10-19 14:11:51 +01:00
53244beab8
add luaossl to lua overlay
...
this means it can be seen by other lua packages (needed for
e.g. luahttp)
2025-10-19 14:11:51 +01:00
3c0f55cd9e
remove outdated cmake patch
2025-10-19 14:11:51 +01:00
d1cd13d5dc
CSR version should be 1 not 3
2025-10-19 14:11:51 +01:00
88dcfc8a05
think
2025-10-19 14:11:51 +01:00
33fdf0d86d
add missing prototype to fetch-freebsd lua glue to make it build
2025-10-19 14:10:13 +01:00
5aa18101b7
shorten bordervm dhcp allocation
...
we use .8 and .10 for static addresses, so start the dhcp range at .12
2025-10-19 14:10:13 +01:00
540c106530
bordervm expose certifix from loachost
2025-10-19 14:10:13 +01:00
4f4aecc8e8
rename openssl => opensslNoThreads, make soft-float if arch 24kc
2025-10-19 14:09:45 +01:00
kurogeek
3f1f7c08d4
Missing repo https://github.com/dfskoll/rp-pppoe is fixed
2025-10-14 14:17:11 +07:00
kurogeek
9621ac51e9
Missing repo https://github.com/kvalo/ath10k-firmware is fixed
2025-10-14 10:14:51 +07:00
cd44343fb7
upgrade luaossl
2025-10-13 19:42:18 +01:00
8b1e871b8a
bordervm forward 9428 (victorialogs) from eth1 to vm hypervisor
2025-10-10 23:42:56 +01:00
bd430f1abd
build luaossl without using luarocks
...
luarocks wants to build cmake, python, curl etc mips packages and I'm
not smart enough to figure out why, but the python doesn't work
2025-10-10 23:39:20 +01:00
77c6552e18
victorialogsend: cope with blank mesage field
2025-10-10 22:43:50 +01:00
a6442c872f
change log shipper interface
...
* it's now a command not a service
* shipper is expected to open() the fifo passed as $LOG_FIFO instead of
getting logs on stdin
logtap determines that remote logs are getting sent when there's a
reader on the fifo it's writing, but opening the fifo as stdin of
s6-tcpclient is too early as it hasn't even connected to the remote
log server at that time
2025-10-09 21:38:49 +01:00
b49beb2c86
make logtap notice sooner when the shipper dies
...
fifo(7) says
Under Linux, opening a FIFO for read and write will succeed both
in blocking and nonblocking mode. POSIX leaves this behavior
undefined. This can be used to open a FIFO for writing while
there are no readers available.
When a process tries to write to a FIFO that is not opened for
read on the other side, the process is sent a SIGPIPE signal.
but ... that doesn't match what I saw, which was that it carried on
writing until the buffer filled and then it blocked or returned -1
(depending on O_NONBLOCK)
Switch to opening O_WRONLY | O_NONBLOCK and retrying periodically
if it fails
2025-10-09 21:16:54 +01:00
12704f1c4a
think
2025-10-08 20:12:34 +01:00
29fbb5461d
send log timestamp parse errors to victorialogs
2025-10-08 20:10:11 +01:00
70786712b3
make victorialogsend ucspi-compatible
2025-10-08 20:10:11 +01:00
be26df4e95
don't send 9 digits of nanosecond to victorialogs
...
It fails to parse. I havent investigated how many digits it will
parse, but let's try 3
2025-10-08 20:10:11 +01:00
71aed767f2
switch logtap to use fifo unstead of af_unix
2025-10-08 20:10:11 +01:00
af6e41db7a
logshippers: add victorialogsend script
...
it's only very lightly tested but it seems to work. the _stream
may be quite wrong, or perhaps that's a local admin decision anyway
2025-10-08 20:10:11 +01:00
947a1c1373
return utc as (values seconds nanos)
...
fractional seconds is just inviting too much fun with floating point
error
2025-10-08 20:10:11 +01:00
59eea64985
add tai64 module to anoia
2025-10-08 20:10:11 +01:00
a343e63231
rename logshipper -> logtap
...
- it matches the executable name
- it doesn't ship anything anyway, it's just plumbing
2025-10-08 20:10:11 +01:00
cacde953cb
don't ask chrony to drop privs, it needs libpcap
2025-10-08 20:10:11 +01:00
9f64eabeb4
rename incz package to logshippers
2025-10-08 20:10:11 +01:00
39fe428e24
bordervm: forward victorialogs connections to host
2025-10-08 20:10:11 +01:00
0147a048fa
bordervm: comment out tang forwarding
2025-10-05 15:00:19 +01:00
ce8a5c1cd3
think
2025-10-05 14:59:15 +01:00
