Hi,
Am Dienstag, den 19.02.2019, 16:37 +0100 schrieb Fabien COELHO
>
> About :
>
> total_percent = total_size ? (int64) ((current_size / MEGABYTES) * 100 /
> (total_size / MEGABYTES)) : 0;
>
> MEGABYTES can be simplified and will enhance precision. ISTM that the
> percent could be a double:
>
> total_percent = total_size ? 100.0 * current_size / total_size : 0.0 ;
>
> and then just let fprintf do the rounding.
Ok, done so.
> I would bother rounding down < 100% to 100, because then you would get
>
> 1560/1492 MB (100\%, X MB/s)
>
> which is kind of silly.
No, we cap the total_size to current_size so you won't see that (but
total_size will potentially gradually increase). pg_basebackup has the
same behaviour.
> I'd still would use more precise than one second precision time so that
> the speed displayed at the beginning is more realistic. eg I got on a
> short test with probably in system cache files:
>
> 188/188 MB (100%, 188 MB/s)
>
> but the reality is that it took 0.126 seconds, and the speed was really
> 1492 MB/s.
Because I implemented I/O throttling for pg_checksums and needed it
there anyway, I've switched it to the instr_time API now. It now updates
the progress 10 times every second.
So now I get
|20/20 MB (100%, 1382 MB/s)
for an empty cluster vs
|20/20 MB (100%, 20 MB/s)
with the previous version of the patch.
New patch attached.
Michael
--
Michael Banck
Projektleiter / Senior Berater
Tel.: +49 2166 9901-171
Fax: +49 2166 9901-100
Email: michael.ba...@credativ.de
credativ GmbH, HRB Mönchengladbach 12080
USt-ID-Nummer: DE204566209
Trompeterallee 108, 41189 Mönchengladbach
Geschäftsführung: Dr. Michael Meskes, Jörg Folz, Sascha Heuer
Unser Umgang mit personenbezogenen Daten unterliegt
folgenden Bestimmungen: https://www.credativ.de/datenschutz
diff --git a/doc/src/sgml/ref/pg_verify_checksums.sgml b/doc/src/sgml/ref/pg_verify_checksums.sgml
index 905b8f1222..b470c19c08 100644
--- a/doc/src/sgml/ref/pg_verify_checksums.sgml
+++ b/doc/src/sgml/ref/pg_verify_checksums.sgml
@@ -80,6 +80,19 @@ PostgreSQL documentation
</varlistentry>
<varlistentry>
+ <term><option>-P</option></term>
+ <term><option>--progress</option></term>
+ <listitem>
+ <para>
+ Enable progress reporting. Turning this on will deliver an approximate
+ progress report during the checksum verification. Sending the
+ <systemitem>SIGUSR1</systemitem> signal will toggle progress reporting
+ on or off during the verification run (not available on Windows).
+ </para>
+ </listitem>
+ </varlistentry>
+
+ <varlistentry>
<term><option>-V</option></term>
<term><option>--version</option></term>
<listitem>
diff --git a/src/bin/pg_verify_checksums/pg_verify_checksums.c b/src/bin/pg_verify_checksums/pg_verify_checksums.c
index 511262ab5f..ca41a3eae7 100644
--- a/src/bin/pg_verify_checksums/pg_verify_checksums.c
+++ b/src/bin/pg_verify_checksums/pg_verify_checksums.c
@@ -10,6 +10,8 @@
#include "postgres_fe.h"
#include <dirent.h>
+#include <signal.h>
+#include <time.h>
#include <sys/stat.h>
#include <unistd.h>
@@ -17,6 +19,7 @@
#include "common/controldata_utils.h"
#include "getopt_long.h"
#include "pg_getopt.h"
+#include "portability/instr_time.h"
#include "storage/bufpage.h"
#include "storage/checksum.h"
#include "storage/checksum_impl.h"
@@ -30,9 +33,18 @@ static ControlFileData *ControlFile;
static char *only_relfilenode = NULL;
static bool verbose = false;
+static bool show_progress = false;
static const char *progname;
+/*
+ * Progress status information.
+ */
+int64 total_size = 0;
+int64 current_size = 0;
+instr_time last_progress_update;
+instr_time scan_started;
+
static void
usage(void)
{
@@ -43,6 +55,7 @@ usage(void)
printf(_(" [-D, --pgdata=]DATADIR data directory\n"));
printf(_(" -v, --verbose output verbose messages\n"));
printf(_(" -r RELFILENODE check only relation with specified relfilenode\n"));
+ printf(_(" -P, --progress show progress information\n"));
printf(_(" -V, --version output version information, then exit\n"));
printf(_(" -?, --help show this help, then exit\n"));
printf(_("\nIf no data directory (DATADIR) is specified, "
@@ -67,6 +80,76 @@ static const char *const skip[] = {
NULL,
};
+static void
+toggle_progress_report(int signum)
+{
+
+ /* we handle SIGUSR1 only, and toggle the value of show_progress */
+ if (signum == SIGUSR1)
+ show_progress = !show_progress;
+
+}
+
+/*
+ * Report current progress status. Parts borrowed from
+ * PostgreSQLs' src/bin/pg_basebackup.c
+ */
+static void
+report_progress(bool force)
+{
+ instr_time now;
+ double elapsed;
+ double total_percent;
+ int64 current_speed;
+
+ char totalstr[32];
+ char currentstr[32];
+ char currspeedstr[32];
+
+ INSTR_TIME_SET_CURRENT(now);
+
+ /* Make sure we report at most once every tenth of a second */
+ if ((INSTR_TIME_GET_MILLISEC(now) -
+ INSTR_TIME_GET_MILLISEC(last_progress_update) < 100) && !force)
+ return;
+
+ /* Save current time */
+ last_progress_update = now;
+
+ /* Elapsed time in milliseconds since start of scan */
+ elapsed = (INSTR_TIME_GET_MILLISEC(now) -
+ INSTR_TIME_GET_MILLISEC(scan_started));
+
+ /* Calculate current percent done */
+ total_percent = total_size ? 100.0 * current_size / total_size : 0.0;
+
+ /* Don't display larger than 100% */
+ if (total_percent > 100)
+ total_percent = 100;
+
+#define MEGABYTES 1048576
+
+ /* The same for total size */
+ if (current_size > total_size)
+ total_size = current_size / MEGABYTES;
+
+ /* Calculate current speed */
+ current_speed = (int64)(current_size / MEGABYTES) / (elapsed / 1000);
+
+ snprintf(totalstr, sizeof(totalstr), INT64_FORMAT,
+ total_size / MEGABYTES);
+ snprintf(currentstr, sizeof(currentstr), INT64_FORMAT,
+ current_size / MEGABYTES);
+ snprintf(currspeedstr, sizeof(currspeedstr), INT64_FORMAT,
+ current_speed);
+
+ fprintf(stderr, "%s/%s MB (%d%%, %s MB/s)",
+ currentstr, totalstr, (int)total_percent, currspeedstr);
+
+ /* Stay on the same line if reporting to a terminal */
+ fprintf(stderr, isatty(fileno(stderr)) ? "\r" : "\n");
+}
+
static bool
skipfile(const char *fn)
{
@@ -117,6 +200,7 @@ scan_file(const char *fn, BlockNumber segmentno)
continue;
csum = pg_checksum_page(buf.data, blockno + segmentno * RELSEG_SIZE);
+ current_size += r;
if (csum != header->pd_checksum)
{
if (ControlFile->data_checksum_version == PG_DATA_CHECKSUM_VERSION)
@@ -124,6 +208,9 @@ scan_file(const char *fn, BlockNumber segmentno)
progname, fn, blockno, csum, header->pd_checksum);
badblocks++;
}
+
+ if (show_progress)
+ report_progress(false);
}
if (verbose)
@@ -133,9 +220,10 @@ scan_file(const char *fn, BlockNumber segmentno)
close(f);
}
-static void
-scan_directory(const char *basedir, const char *subdir)
+static int64
+scan_directory(const char *basedir, const char *subdir, bool sizeonly)
{
+ int64 dirsize = 0;
char path[MAXPGPATH];
DIR *dir;
struct dirent *de;
@@ -167,7 +255,7 @@ scan_directory(const char *basedir, const char *subdir)
if (strncmp(de->d_name,
PG_TEMP_FILES_DIR,
strlen(PG_TEMP_FILES_DIR)) == 0)
- return;
+ continue;
snprintf(fn, sizeof(fn), "%s/%s", path, de->d_name);
if (lstat(fn, &st) < 0)
@@ -214,16 +302,20 @@ scan_directory(const char *basedir, const char *subdir)
/* Relfilenode not to be included */
continue;
- scan_file(fn, segmentno);
+ dirsize += st.st_size;
+
+ if (!sizeonly)
+ scan_file(fn, segmentno);
}
#ifndef WIN32
else if (S_ISDIR(st.st_mode) || S_ISLNK(st.st_mode))
#else
else if (S_ISDIR(st.st_mode) || pgwin32_is_junction(fn))
#endif
- scan_directory(path, de->d_name);
+ dirsize += scan_directory(path, de->d_name, sizeonly);
}
closedir(dir);
+ return dirsize;
}
int
@@ -231,6 +323,7 @@ main(int argc, char *argv[])
{
static struct option long_options[] = {
{"pgdata", required_argument, NULL, 'D'},
+ {"progress", no_argument, NULL, 'P'},
{"verbose", no_argument, NULL, 'v'},
{NULL, 0, NULL, 0}
};
@@ -258,7 +351,7 @@ main(int argc, char *argv[])
}
}
- while ((c = getopt_long(argc, argv, "D:r:v", long_options, &option_index)) != -1)
+ while ((c = getopt_long(argc, argv, "D:r:vP", long_options, &option_index)) != -1)
{
switch (c)
{
@@ -276,6 +369,9 @@ main(int argc, char *argv[])
}
only_relfilenode = pstrdup(optarg);
break;
+ case 'P':
+ show_progress = true;
+ break;
default:
fprintf(stderr, _("Try \"%s --help\" for more information.\n"), progname);
exit(1);
@@ -329,10 +425,45 @@ main(int argc, char *argv[])
exit(1);
}
+
+#ifndef WIN32
+ /*
+ * Assign SIGUSR1 signal handler to toggle progress status information
+ */
+ pqsignal(SIGUSR1, toggle_progress_report);
+#endif
+
+ /*
+ * As progress status information may be requested even after start of
+ * operation, we need to scan the directory tree(s) twice, once to get
+ * the idea how much data we need to scan and finally to do the real
+ * legwork.
+ */
+ total_size = scan_directory(DataDir, "global", true);
+ total_size += scan_directory(DataDir, "base", true);
+ total_size += scan_directory(DataDir, "pg_tblspc", true);
+
+ /*
+ * Remember start time. Required to calculate the current speed in
+ * report_progress()
+ */
+ INSTR_TIME_SET_CURRENT(scan_started);
+
/* Scan all files */
- scan_directory(DataDir, "global");
- scan_directory(DataDir, "base");
- scan_directory(DataDir, "pg_tblspc");
+ scan_directory(DataDir, "global", false);
+ scan_directory(DataDir, "base", false);
+ scan_directory(DataDir, "pg_tblspc", false);
+
+ /*
+ * Done. Move to next line in case progress information was shown.
+ * Otherwise we clutter the summary output.
+ */
+ if (show_progress)
+ {
+ report_progress(true);
+ if (isatty(fileno(stderr)))
+ fprintf(stderr, "\n");
+ }
printf(_("Checksum scan completed\n"));
printf(_("Data checksum version: %d\n"), ControlFile->data_checksum_version);
