On 3/25/22 2:54 PM, Daniel P. Berrangé wrote: > On Fri, Mar 25, 2022 at 02:52:05PM +0100, Claudio Fontana wrote: >> On 3/25/22 2:13 PM, Daniel P. Berrangé wrote: >>> On Fri, Mar 25, 2022 at 01:54:51PM +0100, Claudio Fontana wrote: >>>> currently the only user of virFileWrapperFdNew is the qemu driver; >>>> virsh save is very slow with a default pipe size. >>>> This change improves throughput by ~400% on fast nvme or ramdisk. >>>> >>>> Best value currently measured is 1MB, which happens to be also >>>> the kernel default for the pipe-max-size. >>>> >>>> Signed-off-by: Claudio Fontana <cfontana@xxxxxxx> >>>> --- >>>> >>>> see v2 at >>>> https://listman.redhat.com/archives/libvir-list/2022-March/229423.html >>>> >>>> Changes v2 -> v3: >>>> >>>> * removed reading of max-pipe-size from procfs, >>>> instead make multiple attempts on EPERM with smaller sizes. >>>> In the regular case, this should succeed on the first try. >>>> (Daniel) >>>> >>>> Changes v1 -> v2: >>>> >>>> * removed VIR_FILE_WRAPPER_BIG_PIPE, made the new pipe resizing >>>> unconditional (Michal) >>>> >>>> * moved code to separate functions (Michal) >>>> >>>> * removed ternary op, disliked in libvirt (Michal) >>>> >>>> * added #ifdef __linux__ (Ani Sinha) >>>> >>>> * try smallest value between currently best measured value (1MB) >>>> and the pipe-max-size setting. If pipe-max-size cannot be read, >>>> try kernel default max (1MB). (Daniel) >>>> >>>> >>>> src/util/virfile.c | 49 ++++++++++++++++++++++++++++++++++++++++++++++ >>>> 1 file changed, 49 insertions(+) >>>> >>>> diff --git a/src/util/virfile.c b/src/util/virfile.c >>>> index a04f888e06..876b865974 100644 >>>> --- a/src/util/virfile.c >>>> +++ b/src/util/virfile.c >>>> @@ -201,6 +201,51 @@ struct _virFileWrapperFd { >>>> }; >>>> >>>> #ifndef WIN32 >>>> + >>>> +#ifdef __linux__ >>>> + >>>> +/** >>>> + * virFileWrapperSetPipeSize: >>>> + * @fd: the fd of the pipe >>>> + * >>>> + * Set best pipe size on the passed file descriptor for bulk transfers of data. >>>> + * >>>> + * default pipe size (usually 64K) is generally not suited for large transfers >>>> + * to fast devices. A value of 1MB has been measured to improve virsh save >>>> + * by 400% in ideal conditions. We retry multiple times with smaller sizes >>>> + * on EPERM to account for possible small values of /proc/sys/fs/pipe-max-size. >>>> + * >>>> + * Return value is 0 on success, -1 and errno set on error. >>>> + * OS note: only for linux, on other OS this is a no-op. >>>> + */ >>>> +static int >>>> +virFileWrapperSetPipeSize(int fd) >>>> +{ >>>> + int sz; >>>> + >>>> + for (sz = 1024 * 1024; sz >= 64 * 1024; sz /= 2) { >>>> + int rv = fcntl(fd, F_SETPIPE_SZ, sz); >>>> + if (rv < 0 && errno == EPERM) { >>>> + continue; /* retry with half the size */ >>>> + } >>>> + if (rv < 0) { >>>> + break; >>>> + } >>>> + VIR_INFO("fd %d pipe size adjusted to %d", fd, sz); >>>> + return 0; >>>> + } >>>> + VIR_WARN("failed to set pipe size to %d (errno=%d)", sz, errno); >>>> + return -1; >>>> +} >>>> + >>>> +#else /* !__linux__ */ >>>> +static int virFileWrapperSetPipeSize(int fd) >>>> +{ >>>> + return 0; >>>> +} >>>> +#endif /* !__linux__ */ >>>> + >>>> + >>>> /** >>>> * virFileWrapperFdNew: >>>> * @fd: pointer to fd to wrap >>>> @@ -282,6 +327,10 @@ virFileWrapperFdNew(int *fd, const char *name, unsigned int flags) >>>> >>>> ret->cmd = virCommandNewArgList(iohelper_path, name, NULL); >>>> >>>> + if (virFileWrapperSetPipeSize(pipefd[!output]) < 0) { >>>> + virReportError(VIR_ERR_SYSTEM_ERROR, "%s", _("unable to set pipe size, data transfer might be slow")); >>> >>> Push this into virFileWrapperSetPipeSize instead of the VIR_WARN >>> there, and use virReportSystemError passing in the errno value too. >> >> >> ok, what about also warning on EPERM? In the normal case we should succeed on the first try I think. > > We generally try to avoid any VIR_WARN in cases that we expect to be > still functional. Users tend to complain when they get warnings for INFO? DEBUG? Or nothing at all? Thanks again > these kind of things. I think coping with smaller max size is a normal > situation, so its merely a perf factor, not a functional problem. > > With regards, > Daniel >
