c/mirror_zfs
1
0
Fork 0
mirror of https://git.proxmox.com/git/mirror_zfs.git synced 2026-05-22 10:37:35 +03:00
Code Issues Packages Projects Releases Wiki Activity

Teach zpool scrub to scrub only blocks in error log

Browse Source 
Added a flag '-e' in zpool scrub to scrub only blocks in error log. A
user can pause, resume and cancel the error scrub by passing additional
command line arguments -p -s just like a regular scrub. This involves
adding a new flag, creating new libzfs interfaces, a new ioctl, and the
actual iteration and read-issuing logic. Error scrubbing is executed in
multiple txg to make sure pool performance is not affected.

Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Tony Hutter <hutter2@llnl.gov>
Co-authored-by: TulsiJain tulsi.jain@delphix.com
Signed-off-by: George Amanakis <gamanakis@gmail.com>
Closes #8995
Closes #12355
This commit is contained in:
George Amanakis
2021-12-17 21:35:28 +01:00
committed by Brian Behlendorf
parent e34e15ed6d
commit 482eeef804
29 changed files with 1602 additions and 71 deletions
Download Patch File Download Diff File Expand all files Collapse all files
cmd/zpool/zpool_main.c
+101 -10
View File
@@ -401,7 +401,7 @@ get_usage(zpool_help_t idx)
return (gettext("\tinitialize [-c | -s | -u] [-w] <pool> "
"[<device> ...]\n"));
case HELP_SCRUB:
return (gettext("\tscrub [-s | -p] [-w] <pool> ...\n"));
return (gettext("\tscrub [-s | -p] [-w] [-e] <pool> ...\n"));
case HELP_RESILVER:
return (gettext("\tresilver <pool> ...\n"));
case HELP_TRIM:
@@ -7309,8 +7309,9 @@ wait_callback(zpool_handle_t *zhp, void *data)
}
/*
* zpool scrub [-s | -p] [-w] <pool> ...
* zpool scrub [-s | -p] [-w] [-e] <pool> ...
*
* -e Only scrub blocks in the error log.
* -s Stop. Stops any in-progress scrub.
* -p Pause. Pause in-progress scrub.
* -w Wait. Blocks until scrub has completed.
@@ -7326,14 +7327,21 @@ zpool_do_scrub(int argc, char **argv)
cb.cb_type = POOL_SCAN_SCRUB;
cb.cb_scrub_cmd = POOL_SCRUB_NORMAL;
boolean_t is_error_scrub = B_FALSE;
boolean_t is_pause = B_FALSE;
boolean_t is_stop = B_FALSE;
/* check options */
while ((c = getopt(argc, argv, "spw")) != -1) {
while ((c = getopt(argc, argv, "spwe")) != -1) {
switch (c) {
case 'e':
is_error_scrub = B_TRUE;
break;
case 's':
cb.cb_type = POOL_SCAN_NONE;
is_stop = B_TRUE;
break;
case 'p':
cb.cb_scrub_cmd = POOL_SCRUB_PAUSE;
is_pause = B_TRUE;
break;
case 'w':
wait = B_TRUE;
@@ -7345,11 +7353,21 @@ zpool_do_scrub(int argc, char **argv)
}
}
if (cb.cb_type == POOL_SCAN_NONE &&
cb.cb_scrub_cmd == POOL_SCRUB_PAUSE) {
(void) fprintf(stderr, gettext("invalid option combination: "
"-s and -p are mutually exclusive\n"));
if (is_pause && is_stop) {
(void) fprintf(stderr, gettext("invalid option "
"combination :-s and -p are mutually exclusive\n"));
usage(B_FALSE);
} else {
if (is_error_scrub)
cb.cb_type = POOL_SCAN_ERRORSCRUB;
if (is_pause) {
cb.cb_scrub_cmd = POOL_SCRUB_PAUSE;
} else if (is_stop) {
cb.cb_type = POOL_SCAN_NONE;
} else {
cb.cb_scrub_cmd = POOL_SCRUB_NORMAL;
}
}
if (wait && (cb.cb_type == POOL_SCAN_NONE ||
@@ -7573,6 +7591,70 @@ secs_to_dhms(uint64_t total, char *buf)
}
}
/*
* Print out detailed error scrub status.
*/
static void
print_err_scrub_status(pool_scan_stat_t *ps)
{
time_t start, end, pause;
uint64_t total_secs_left;
uint64_t secs_left, mins_left, hours_left, days_left;
uint64_t examined, to_be_examined;
if (ps == NULL || ps->pss_error_scrub_func != POOL_SCAN_ERRORSCRUB) {
return;
}
(void) printf(gettext(" scrub: "));
start = ps->pss_error_scrub_start;
end = ps->pss_error_scrub_end;
pause = ps->pss_pass_error_scrub_pause;
examined = ps->pss_error_scrub_examined;
to_be_examined = ps->pss_error_scrub_to_be_examined;
assert(ps->pss_error_scrub_func == POOL_SCAN_ERRORSCRUB);
if (ps->pss_error_scrub_state == DSS_FINISHED) {
total_secs_left = end - start;
days_left = total_secs_left / 60 / 60 / 24;
hours_left = (total_secs_left / 60 / 60) % 24;
mins_left = (total_secs_left / 60) % 60;
secs_left = (total_secs_left % 60);
(void) printf(gettext("scrubbed %llu error blocks in %llu days "
"%02llu:%02llu:%02llu on %s"), (u_longlong_t)examined,
(u_longlong_t)days_left, (u_longlong_t)hours_left,
(u_longlong_t)mins_left, (u_longlong_t)secs_left,
ctime(&end));
return;
} else if (ps->pss_error_scrub_state == DSS_CANCELED) {
(void) printf(gettext("error scrub canceled on %s"),
ctime(&end));
return;
}
assert(ps->pss_error_scrub_state == DSS_ERRORSCRUBBING);
/* Error scrub is in progress. */
if (pause == 0) {
(void) printf(gettext("error scrub in progress since %s"),
ctime(&start));
} else {
(void) printf(gettext("error scrub paused since %s"),
ctime(&pause));
(void) printf(gettext("\terror scrub started on %s"),
ctime(&start));
}
double fraction_done = (double)examined / (to_be_examined + examined);
(void) printf(gettext("\t%.2f%% done, issued I/O for %llu error"
" blocks"), 100 * fraction_done, (u_longlong_t)examined);
(void) printf("\n");
}
/*
* Print out detailed scrub status.
*/
@@ -7909,10 +7991,12 @@ print_scan_status(zpool_handle_t *zhp, nvlist_t *nvroot)
{
uint64_t rebuild_end_time = 0, resilver_end_time = 0;
boolean_t have_resilver = B_FALSE, have_scrub = B_FALSE;
boolean_t have_errorscrub = B_FALSE;
boolean_t active_resilver = B_FALSE;
pool_checkpoint_stat_t *pcs = NULL;
pool_scan_stat_t *ps = NULL;
uint_t c;
time_t scrub_start = 0, errorscrub_start = 0;
if (nvlist_lookup_uint64_array(nvroot, ZPOOL_CONFIG_SCAN_STATS,
(uint64_t **)&ps, &c) == 0) {
@@ -7921,16 +8005,23 @@ print_scan_status(zpool_handle_t *zhp, nvlist_t *nvroot)
active_resilver = (ps->pss_state == DSS_SCANNING);
}
have_resilver = (ps->pss_func == POOL_SCAN_RESILVER);
have_scrub = (ps->pss_func == POOL_SCAN_SCRUB);
scrub_start = ps->pss_start_time;
have_errorscrub = (ps->pss_error_scrub_func ==
POOL_SCAN_ERRORSCRUB);
errorscrub_start = ps->pss_error_scrub_start;
}
boolean_t active_rebuild = check_rebuilding(nvroot, &rebuild_end_time);
boolean_t have_rebuild = (active_rebuild || (rebuild_end_time > 0));
/* Always print the scrub status when available. */
if (have_scrub)
if (have_scrub && scrub_start > errorscrub_start)
print_scan_scrub_resilver_status(ps);
else if (have_errorscrub && errorscrub_start >= scrub_start)
print_err_scrub_status(ps);
/*
* When there is an active resilver or rebuild print its status.