From fb2b86015a9f3e2e59ddbb59e18cda98363cbd7b Mon Sep 17 00:00:00 2001 From: Michael Paquier Date: Tue, 20 Jul 2021 12:12:51 +0900 Subject: [PATCH] Fix some issues with WAL segment opening for pg_receivewal --compress The logic handling the opening of new WAL segments was fuzzy when using --compress if a partial, non-compressed, segment with the same base name existed in the repository storing those files. In this case, using --compress would cause the code to first check for the existence and the size of a non-compressed segment, followed by the opening of a new compressed, partial, segment. The code was accidentally working correctly on most platforms as the buildfarm has proved, except bowerbird where gzflush() could fail in this code path. It is wrong anyway to take the code path used pre-padding when creating a new partial, non-compressed, segment, so let's fix it. Note that this issue exists when users mix successive runs of pg_receivewal with or without compression, as discovered with the tests introduced by ffc9dda. While on it, this refactors the code so as code paths that need to know about the ".gz" suffix are down from four to one in walmethods.c, easing a bit the introduction of new compression methods. This addresses a second issue where log messages generated for an unexpected failure would not show the compressed segment name involved, which was confusing, printing instead the name of the non-compressed equivalent. Reported-by: Georgios Kokolatos Discussion: https://postgr.es/m/YPDLz2x3o1aX2wRh@paquier.xyz Backpatch-through: 10 --- src/bin/pg_basebackup/receivelog.c | 13 +++-- src/bin/pg_basebackup/walmethods.c | 79 +++++++++++++++++++++++------- src/bin/pg_basebackup/walmethods.h | 9 ++++ 3 files changed, 78 insertions(+), 23 deletions(-) diff --git a/src/bin/pg_basebackup/receivelog.c b/src/bin/pg_basebackup/receivelog.c index 62a342f77c..103c28b50b 100644 --- a/src/bin/pg_basebackup/receivelog.c +++ b/src/bin/pg_basebackup/receivelog.c @@ -89,26 +89,29 @@ static bool open_walfile(StreamCtl *stream, XLogRecPtr startpoint) { Walfile *f; - char fn[MAXPGPATH]; + char *fn; ssize_t size; XLogSegNo segno; XLByteToSeg(startpoint, segno, WalSegSz); XLogFileName(current_walfile_name, stream->timeline, segno, WalSegSz); - snprintf(fn, sizeof(fn), "%s%s", current_walfile_name, - stream->partial_suffix ? stream->partial_suffix : ""); + /* Note that this considers the compression used if necessary */ + fn = stream->walmethod->get_file_name(current_walfile_name, + stream->partial_suffix); /* * When streaming to files, if an existing file exists we verify that it's * either empty (just created), or a complete WalSegSz segment (in which * case it has been created and padded). Anything else indicates a corrupt - * file. + * file. Compressed files have no need for padding, so just ignore this + * case. * * When streaming to tar, no file with this name will exist before, so we * never have to verify a size. */ - if (stream->walmethod->existsfile(fn)) + if (stream->walmethod->compression() == 0 && + stream->walmethod->existsfile(fn)) { size = stream->walmethod->get_file_size(fn); if (size < 0) diff --git a/src/bin/pg_basebackup/walmethods.c b/src/bin/pg_basebackup/walmethods.c index ecff08740c..ed272bfa55 100644 --- a/src/bin/pg_basebackup/walmethods.c +++ b/src/bin/pg_basebackup/walmethods.c @@ -68,20 +68,32 @@ dir_getlasterror(void) return strerror(errno); } +static char * +dir_get_file_name(const char *pathname, const char *temp_suffix) +{ + char *filename = pg_malloc0(MAXPGPATH * sizeof(char)); + + snprintf(filename, MAXPGPATH, "%s%s%s", + pathname, dir_data->compression > 0 ? ".gz" : "", + temp_suffix ? temp_suffix : ""); + + return filename; +} + static Walfile dir_open_for_write(const char *pathname, const char *temp_suffix, size_t pad_to_size) { static char tmppath[MAXPGPATH]; + char *filename; int fd; DirectoryMethodFile *f; #ifdef HAVE_LIBZ gzFile gzfp = NULL; #endif - snprintf(tmppath, sizeof(tmppath), "%s/%s%s%s", - dir_data->basedir, pathname, - dir_data->compression > 0 ? ".gz" : "", - temp_suffix ? temp_suffix : ""); + filename = dir_get_file_name(pathname, temp_suffix); + snprintf(tmppath, sizeof(tmppath), "%s/%s", + dir_data->basedir, filename); /* * Open a file for non-compressed as well as compressed files. Tracking @@ -232,26 +244,31 @@ dir_close(Walfile f, WalCloseMethod method) /* Build path to the current version of the file */ if (method == CLOSE_NORMAL && df->temp_suffix) { + char *filename; + char *filename2; + /* * If we have a temp prefix, normal operation is to rename the * file. */ - snprintf(tmppath, sizeof(tmppath), "%s/%s%s%s", - dir_data->basedir, df->pathname, - dir_data->compression > 0 ? ".gz" : "", - df->temp_suffix); - snprintf(tmppath2, sizeof(tmppath2), "%s/%s%s", - dir_data->basedir, df->pathname, - dir_data->compression > 0 ? ".gz" : ""); + filename = dir_get_file_name(df->pathname, df->temp_suffix); + snprintf(tmppath, sizeof(tmppath), "%s/%s", + dir_data->basedir, filename); + + /* permanent name, so no need for the prefix */ + filename2 = dir_get_file_name(df->pathname, NULL); + snprintf(tmppath2, sizeof(tmppath2), "%s/%s", + dir_data->basedir, filename2); r = durable_rename(tmppath, tmppath2); } else if (method == CLOSE_UNLINK) { + char *filename; + /* Unlink the file once it's closed */ - snprintf(tmppath, sizeof(tmppath), "%s/%s%s%s", - dir_data->basedir, df->pathname, - dir_data->compression > 0 ? ".gz" : "", - df->temp_suffix ? df->temp_suffix : ""); + filename = dir_get_file_name(df->pathname, df->temp_suffix); + snprintf(tmppath, sizeof(tmppath), "%s/%s", + dir_data->basedir, filename); r = unlink(tmppath); } else @@ -313,6 +330,12 @@ dir_get_file_size(const char *pathname) return statbuf.st_size; } +static int +dir_compression(void) +{ + return dir_data->compression; +} + static bool dir_existsfile(const char *pathname) { @@ -355,6 +378,8 @@ CreateWalDirectoryMethod(const char *basedir, int compression, bool sync) method->write = dir_write; method->get_current_pos = dir_get_current_pos; method->get_file_size = dir_get_file_size; + method->get_file_name = dir_get_file_name; + method->compression = dir_compression; method->close = dir_close; method->sync = dir_sync; method->existsfile = dir_existsfile; @@ -527,11 +552,22 @@ tar_write_padding_data(TarMethodFile *f, size_t bytes) return true; } +static char * +tar_get_file_name(const char *pathname, const char *temp_suffix) +{ + char *filename = pg_malloc0(MAXPGPATH * sizeof(char)); + + snprintf(filename, MAXPGPATH, "%s%s", + pathname, temp_suffix ? temp_suffix : ""); + + return filename; +} + static Walfile tar_open_for_write(const char *pathname, const char *temp_suffix, size_t pad_to_size) { int save_errno; - static char tmppath[MAXPGPATH]; + char *tmppath; tar_clear_error(); @@ -583,8 +619,7 @@ tar_open_for_write(const char *pathname, const char *temp_suffix, size_t pad_to_ tar_data->currentfile = pg_malloc0(sizeof(TarMethodFile)); - snprintf(tmppath, sizeof(tmppath), "%s%s", - pathname, temp_suffix ? temp_suffix : ""); + tmppath = tar_get_file_name(pathname, temp_suffix); /* Create a header with size set to 0 - we will fill out the size on close */ if (tarCreateHeader(tar_data->currentfile->header, tmppath, NULL, 0, S_IRUSR | S_IWUSR, 0, 0, time(NULL)) != TAR_OK) @@ -685,6 +720,12 @@ tar_get_file_size(const char *pathname) return -1; } +static int +tar_compression(void) +{ + return tar_data->compression; +} + static off_t tar_get_current_pos(Walfile f) { @@ -989,6 +1030,8 @@ CreateWalTarMethod(const char *tarbase, int compression, bool sync) method->write = tar_write; method->get_current_pos = tar_get_current_pos; method->get_file_size = tar_get_file_size; + method->get_file_name = tar_get_file_name; + method->compression = tar_compression; method->close = tar_close; method->sync = tar_sync; method->existsfile = tar_existsfile; diff --git a/src/bin/pg_basebackup/walmethods.h b/src/bin/pg_basebackup/walmethods.h index 9a661c673c..f9bd59d0cd 100644 --- a/src/bin/pg_basebackup/walmethods.h +++ b/src/bin/pg_basebackup/walmethods.h @@ -52,6 +52,15 @@ struct WalWriteMethod /* Return the size of a file, or -1 on failure. */ ssize_t (*get_file_size) (const char *pathname); + /* + * Return the name of the current file to work on, without the base + * directory. This is useful for logging. + */ + char *(*get_file_name) (const char *pathname, const char *temp_suffix); + + /* Return the level of compression */ + int (*compression) (void); + /* * Write count number of bytes to the file, and return the number of bytes * actually written or -1 for error.