Tried to debug it a bit more, with the attached test program. It has
printed the first message from each 4-hour chunk of messages (mentioned
as "<message 1>" in the original report), but nothing else -- though
there are others that match ("<message 2>" and so on), including the
ones written by the same process as those first-in-group messages. This
behaviour is different from what journalctl does, but also erroneous. In
both cases (with journalctl and with this test program), it goes back
too far, to the same record.
For another OBJECT_ID (for which journalctl works fine, and with less
repetitive messages), the test program has printed the logs seemingly
correctly, without skipping anything.
#include <systemd/sd-journal.h>
#include <stdio.h>
int read_journal (sd_journal *j) {
const char *d;
size_t l;
if (sd_journal_add_match(j, "OBJECT_ID=50482", 0) < 0) {
perror("Failed to add match");
return 1;
}
SD_JOURNAL_FOREACH_BACKWARDS(j) {
puts("--");
SD_JOURNAL_FOREACH_DATA(j, d, l) {
printf("%.*s\n", (int) l, d);
}
}
return 0;
}
int main () {
int ret;
sd_journal *j;
if (sd_journal_open(&j, 0) < 0) {
perror("Failed to open journal");
return 1;
}
ret = read_journal(j);
sd_journal_close(j);
return ret;
}
_______________________________________________
Pkg-systemd-maintainers mailing list
Pkg-systemd-maintainers@lists.alioth.debian.org
http://lists.alioth.debian.org/cgi-bin/mailman/listinfo/pkg-systemd-maintainers
