Hi, Well I didn't know that a few days can pass as fast :-) I have attached the patch to this mail. Until now the patch has only been shortly tested on a loop device, so it might contain bugs and destroy your data. If you decide to apply it, please change the default values to the ones you find the most appropriate. Thanks in advance, Bye, David Arendt On 03/15/10 16:58, Ryusuke Konishi wrote: > Hi, > On Mon, 15 Mar 2010 00:03:45 +0100, David Arendt wrote: > >> Hi, >> >> I am posting this again to the correct mailing list as I cc'ed it to the >> old inactive one. >> >> Maybe I am understanding something wrong, but if I would use the count >> of reclaimed segments, how could I determine if one cleaning pass has >> finished as I don't know in advance how many segments could be reclaimed ? >> > For example, how about this? > > nmax = (number of segments) - (number of clean segments) > nblk = (max_clean_segments - (number of clean segments)) * > (number of blocks per segment) > > * If (number of clean segments) < min_clean_segments, then start reclamation > * Try to reclaim nmax segments (at a maximum). > * When the cleaner found and freed nblk blocks during the > reclamation, then end one cleaning pass. > > >> Another approach would be not basing cleaning on a whole cleaning pass >> but instead creating these addtional configfile options: >> >> # start cleaning if less than 100 free segments >> min_clean_segments 100 >> >> # stop cleaning if more than 200 free segments >> max_clean_segments 200 >> >> # check free space once an hour >> segment_check_interval 3600 >> >> Basically in this example if less than 800mb are free cleaner is run >> until 1600mb are free. If min_clean_segments is 0, the cleaner would do >> normal operation. >> > The first two parameters look Ok. > (I've already referred to these in the above example.) > > We may well be able to make segment_check_interval more frequent. > or do you have something in mind? > > Do you mean interval of cleaning passes ? > > >> For this solution only changes in configfile loading and >> nilfs_cleanerd_clean_loop would be necessary which would lower the risk >> of introducing new bugs. >> >> If this solution is ok for you, I will implement it this way and send >> you the patch in a few days. Also tell me if the names I have choosen >> for the options are ok for you or if you would prefer other ones. >> > The option names look fine to me. > Or should we use percentage for them? > (number of segments is device dependent) > > Is there anything else that isn't clear? > > >> Thanks in advance >> Bye, >> David Arendt >> > Thanks, > Ryusuke Konishi > > >> On 03/14/10 15:28, Ryusuke Konishi wrote: >> >>> Hi, >>> On Sun, 14 Mar 2010 14:00:19 +0100, admin@xxxxxxxxx wrote: >>> >>> >>>> Hi, >>>> >>>> I will try to implement this myself then. Concerning the >>>> nilfs_cleanerd_select segments function I was unclear in my post. In >>>> fact I did not mean the return value but the first element from the >>>> segnums array. >>>> >>>> >>> Ok. So you thought of determining termination of one cleaning pass by >>> the segment number stored preliminarily. >>> >>> Why not just use count of processed (i.e. reclaimed) segments? >>> >>> Note that it's not guranteed that segments are selected in the order >>> of segment number though this premise looks almost right. >>> >>> It depends on the behavior of segment allocator and the current >>> "Select-oldest" algorithm used behind >>> nilfs_cleanerd_select_segments(). Nilfs log writer occasionally >>> behaves differently and disturbs this order. >>> >>> I think you can ignore the exceptional behavior of the segment >>> allocator, and rotate target segments with skipping free or mostly >>> in-use ones. In that case, nilfs_cleanerd_select_segments() should be >>> modified to select segments in the order of segment number. >>> >>> Cheers, >>> Ryusuke Konishi >>> >>> >>> >> -- >> To unsubscribe from this list: send the line "unsubscribe linux-nilfs" in >> the body of a message to majordomo@xxxxxxxxxxxxxxx >> More majordomo info at http://vger.kernel.org/majordomo-info.html >>
diff -ur nilfs2-utils.orig/man/nilfs_cleanerd.conf.5 nilfs2-utils/man/nilfs_cleanerd.conf.5 --- nilfs2-utils.orig/man/nilfs_cleanerd.conf.5 2010-03-14 15:11:30.916690347 +0100 +++ nilfs2-utils/man/nilfs_cleanerd.conf.5 2010-03-15 22:15:58.320507660 +0100 @@ -25,6 +25,23 @@ and their blocks whose duration time is less than the value. The default value is 3600, meaning one hour. .TP +.B min_clean_segments +Specify the minimum number of clean segments. A value of 0 means +normal cleaner operation. A value greater than 0 means pause cleaning +until less than min_clean_segments are available. The default value +is 100. +.TP +.B max_clean_segments +Specify the maximum number of clean segments. If min_clean_segments is +0, this value is ignored. If more than max_clean_segments are available +cleaning is paused until less than min_clean_segments are available. +The default value is 200. +.TP +.B clean_check_interval +Specify the interval to wait between checks of min_clean_segments. +If min_clean_segments is 0, this value is ignored. +The default value is 60. +.TP .B selection_policy Specify the GC policy. At present, only the `\fBtimestamp\fP' policy, which reclaims segments in order from oldest to newest, is support. diff -ur nilfs2-utils.orig/sbin/cleanerd/cldconfig.c nilfs2-utils/sbin/cleanerd/cldconfig.c --- nilfs2-utils.orig/sbin/cleanerd/cldconfig.c 2010-03-14 15:11:30.916690347 +0100 +++ nilfs2-utils/sbin/cleanerd/cldconfig.c 2010-03-15 20:20:22.364435601 +0100 @@ -90,6 +90,87 @@ return 0; } +static int +nilfs_cldconfig_handle_min_clean_segments(struct nilfs_cldconfig *config, + char **tokens, size_t ntoks) +{ + __u64 n; + char *endptr; + + if (check_tokens(tokens, ntoks, 2, 2) < 0) + return 0; + + errno = 0; + n = strtoull(tokens[1], &endptr, 10); + if (*endptr != '\0') { + syslog(LOG_WARNING, "%s: %s: not a number", + tokens[0], tokens[1]); + return 0; + } + if ((n == ULLONG_MAX) && (errno == ERANGE)) { + syslog(LOG_WARNING, "%s: %s: number too large", + tokens[0], tokens[1]); + return 0; + } + + config->cf_min_clean_segments = n; + return 0; +} + +static int +nilfs_cldconfig_handle_max_clean_segments(struct nilfs_cldconfig *config, + char **tokens, size_t ntoks) +{ + __u64 n; + char *endptr; + + if (check_tokens(tokens, ntoks, 2, 2) < 0) + return 0; + + errno = 0; + n = strtoull(tokens[1], &endptr, 10); + if (*endptr != '\0') { + syslog(LOG_WARNING, "%s: %s: not a number", + tokens[0], tokens[1]); + return 0; + } + if ((n == ULLONG_MAX) && (errno == ERANGE)) { + syslog(LOG_WARNING, "%s: %s: number too large", + tokens[0], tokens[1]); + return 0; + } + + config->cf_max_clean_segments = n; + return 0; +} + +static int +nilfs_cldconfig_handle_clean_check_interval(struct nilfs_cldconfig *config, + char **tokens, size_t ntoks) +{ + time_t period; + char *endptr; + + if (check_tokens(tokens, ntoks, 2, 2) < 0) + return 0; + + errno = 0; + period = strtoul(tokens[1], &endptr, 10); + if (*endptr != '\0') { + syslog(LOG_WARNING, "%s: %s: not a number", + tokens[0], tokens[1]); + return 0; + } + if ((period == ULONG_MAX) && (errno == ERANGE)) { + syslog(LOG_WARNING, "%s: %s: number too large", + tokens[0], tokens[1]); + return 0; + } + + config->cf_clean_check_interval = period; + return 0; +} + static unsigned long long nilfs_cldconfig_selection_policy_timestamp(const struct nilfs_suinfo *si) { @@ -277,6 +358,9 @@ static const struct nilfs_cldconfig_keyword nilfs_cldconfig_keyword_table[] = { {"protection_period", nilfs_cldconfig_handle_protection_period}, + {"min_clean_segments", nilfs_cldconfig_handle_min_clean_segments}, + {"max_clean_segments", nilfs_cldconfig_handle_max_clean_segments}, + {"clean_check_interval",nilfs_cldconfig_handle_clean_check_interval}, {"selection_policy", nilfs_cldconfig_handle_selection_policy}, {"nsegments_per_clean", nilfs_cldconfig_handle_nsegments_per_clean}, {"cleaning_interval", nilfs_cldconfig_handle_cleaning_interval}, @@ -313,6 +397,9 @@ config->cf_selection_policy.p_threshold = NILFS_CLDCONFIG_SELECTION_POLICY_THRESHOLD; config->cf_protection_period = NILFS_CLDCONFIG_PROTECTION_PERIOD; + config->cf_min_clean_segments = NILFS_CLDCONFIG_MIN_CLEAN_SEGMENTS; + config->cf_max_clean_segments = NILFS_CLDCONFIG_MAX_CLEAN_SEGMENTS; + config->cf_clean_check_interval = NILFS_CLDCONFIG_CLEAN_CHECK_INTERVAL; config->cf_nsegments_per_clean = NILFS_CLDCONFIG_NSEGMENTS_PER_CLEAN; config->cf_cleaning_interval = NILFS_CLDCONFIG_CLEANING_INTERVAL; config->cf_retry_interval = NILFS_CLDCONFIG_RETRY_INTERVAL; diff -ur nilfs2-utils.orig/sbin/cleanerd/cldconfig.h nilfs2-utils/sbin/cleanerd/cldconfig.h --- nilfs2-utils.orig/sbin/cleanerd/cldconfig.h 2010-03-14 15:11:30.916690347 +0100 +++ nilfs2-utils/sbin/cleanerd/cldconfig.h 2010-03-15 21:58:16.208614224 +0100 @@ -42,6 +42,9 @@ * struct nilfs_cldconfig - * @cf_selection_policy: * @cf_protection_period: + * @cf_min_clean_segments: + * @cf_max_clean_segments: + * @cf_clean_check_interval: * @cf_nsegments_per_clean * @cf_cleaning_interval: * @cf_use_mmap: @@ -50,6 +53,9 @@ struct nilfs_cldconfig { struct nilfs_selection_policy cf_selection_policy; time_t cf_protection_period; + __u64 cf_min_clean_segments; + __u64 cf_max_clean_segments; + time_t cf_clean_check_interval; int cf_nsegments_per_clean; time_t cf_cleaning_interval; time_t cf_retry_interval; @@ -61,6 +67,9 @@ nilfs_cldconfig_selection_policy_timestamp #define NILFS_CLDCONFIG_SELECTION_POLICY_THRESHOLD 0 #define NILFS_CLDCONFIG_PROTECTION_PERIOD 3600 +#define NILFS_CLDCONFIG_MIN_CLEAN_SEGMENTS 100 +#define NILFS_CLDCONFIG_MAX_CLEAN_SEGMENTS 200 +#define NILFS_CLDCONFIG_CLEAN_CHECK_INTERVAL 60 #define NILFS_CLDCONFIG_NSEGMENTS_PER_CLEAN 2 #define NILFS_CLDCONFIG_CLEANING_INTERVAL 5 #define NILFS_CLDCONFIG_RETRY_INTERVAL 60 diff -ur nilfs2-utils.orig/sbin/cleanerd/cleanerd.c nilfs2-utils/sbin/cleanerd/cleanerd.c --- nilfs2-utils.orig/sbin/cleanerd/cleanerd.c 2010-03-14 15:11:30.916690347 +0100 +++ nilfs2-utils/sbin/cleanerd/cleanerd.c 2010-03-15 21:41:32.956273102 +0100 @@ -1175,6 +1175,7 @@ struct timespec timeout; sigset_t sigset; int ns, ret; + int sleeping = 1; sigemptyset(&sigset); if (sigprocmask(SIG_SETMASK, &sigset, NULL) < 0) { @@ -1201,6 +1202,12 @@ cleanerd->c_running = 1; cleanerd->c_ncleansegs = cleanerd->c_config.cf_nsegments_per_clean; + if (cleanerd->c_config.cf_min_clean_segments > 0) { + syslog(LOG_INFO, "cleaner paused"); + timeout.tv_sec = cleanerd->c_config.cf_clean_check_interval; + timeout.tv_nsec = 0; + } + while (1) { if (sigprocmask(SIG_BLOCK, &sigset, NULL) < 0) { syslog(LOG_ERR, "cannot set signal mask: %m"); @@ -1220,10 +1227,32 @@ syslog(LOG_ERR, "cannot get segment usage stat: %m"); return -1; } + + if (cleanerd->c_config.cf_min_clean_segments > 0) { + if (sleeping) { + if (sustat.ss_ncleansegs < cleanerd->c_config.cf_min_clean_segments) { + syslog(LOG_INFO, "cleaner resumed"); + sleeping = 0; + } + else + goto sleep; + } + else { + if (sustat.ss_ncleansegs > cleanerd->c_config.cf_max_clean_segments) { + syslog(LOG_INFO, "cleaner paused"); + sleeping = 1; + timeout.tv_sec = cleanerd->c_config.cf_clean_check_interval; + timeout.tv_nsec = 0; + goto sleep; + } + } + } + if (sustat.ss_nongc_ctime != prev_nongc_ctime) { cleanerd->c_running = 1; prev_nongc_ctime = sustat.ss_nongc_ctime; } + if (!cleanerd->c_running) goto sleep; diff -ur nilfs2-utils.orig/sbin/cleanerd/nilfs_cleanerd.conf nilfs2-utils/sbin/cleanerd/nilfs_cleanerd.conf --- nilfs2-utils.orig/sbin/cleanerd/nilfs_cleanerd.conf 2010-03-14 15:11:30.916690347 +0100 +++ nilfs2-utils/sbin/cleanerd/nilfs_cleanerd.conf 2010-03-15 21:44:02.995587453 +0100 @@ -7,6 +7,17 @@ # Protection period in second. protection_period 3600 +# Minium number of clean segements +# 0 = normal cleaner behaviour +# >0 = start cleaning if less segments are available +min_clean_segments 100 + +# Maximum number of clean segments +max_clean_segments 200 + +# Clean segment check interval in seconds +clean_check_interval 60 + # Segment selection policy. # In NILFS version 2.0.0, only the timestamp policy is supported. selection_policy timestamp # timestamp in ascend order