August 2020 - Libguestfs - libguestfs List Archives

[PATCH] daemon: inspect_fs_windows: Handle parted errors
by Sam Eiderman 22 Sep '20

22 Sep '20

By creating an empty disk and using it as the first disk of the vm (i.e. /dev/sda, /dev/sdb{1,2} contains the windows fses) we change the iteration order of the disks. This causes inspect_os() to fail since Parted returns a Unix_error if the device does not contain any partitions - fix this by handling this Unix_error. Signed-off-by: Sam Eiderman <sameid(a)google.com> --- daemon/inspect_fs_windows.ml | 11 ++++++++--- 1 file changed, 8 insertions(+), 3 deletions(-) diff --git a/daemon/inspect_fs_windows.ml b/daemon/inspect_fs_windows.ml index c4a05bc38..bc6b98b60 100644 --- a/daemon/inspect_fs_windows.ml +++ b/daemon/inspect_fs_windows.ml @@ -365,8 +365,10 @@ and map_registry_disk_blob_mbr devices blob = let device = List.find ( fun dev -> - Parted.part_get_parttype dev = "msdos" && - pread dev 4 0x01b8 = diskid + try + Parted.part_get_parttype dev = "msdos" && + pread dev 4 0x01b8 = diskid + with Unix.Unix_error (Unix.EINVAL, _, _) -> false ) devices in (* Next 8 bytes are the offset of the partition in bytes(!) given as @@ -402,7 +404,10 @@ and map_registry_disk_blob_gpt partitions blob = fun part -> let partnum = Devsparts.part_to_partnum part in let device = Devsparts.part_to_dev part in - let typ = Parted.part_get_parttype device in + let typ = + try + Parted.part_get_parttype device + with Unix.Unix_error (Unix.EINVAL, _, _) -> "unknown" in if typ <> "gpt" then false else ( let guid = Parted.part_get_gpt_guid device partnum in -- 2.27.0.212.ge8ba1cc988-goog

2 3

[PATCH v2v] Add ALT support
by Richard W.M. Jones 10 Sep '20

10 Sep '20

Patch supplied by Mikhail Gordeev, posting for review. I have compile tested it and checked the code and it looks all fine to me, so ACK from my point of view. I did not actually run it because I don't have an ALT Linux install, but it doesn't seem as if it would affect any other distro. Rich.

3 5

[PATCH nbdkit NOT WORKING 0/2] vddk: Relax threading model.
by Richard W.M. Jones 08 Sep '20

08 Sep '20

I believe this roughly implements Nir's proposal here: https://www.redhat.com/archives/libguestfs/2020-August/msg00028.html Unfortunately it doesn't work for me. It actually slows things down quite a lot, for reasons I don't understand. Note the adjustment of the pool-max parameter and how it affects the total time. The results are quite reproducible. $ ./nbdkit -r -U - vddk libdir=~/tmp/vddk-7.0.0/vmware-vix-disklib-distrib user=root password=*** server=*** thumbprint=*** vm=moref=3 file='[datastore1] Fedora 28/Fedora 28.vmdk' pool-max=1 --run 'time qemu-img convert -p -W -m 16 $nbd /var/tmp/out' (100.00/100%) real 1m8.031s user 0m0.112s sys 0m1.560s $ ./nbdkit -r -U - vddk libdir=~/tmp/vddk-7.0.0/vmware-vix-disklib-distrib user=root password=*** server=*** thumbprint=*** vm=moref=3 file='[datastore1] Fedora 28/Fedora 28.vmdk' pool-max=8 --run 'time qemu-img convert -p -W -m 16 $nbd /var/tmp/out' (100.00/100%) real 1m27.790s user 0m0.099s sys 0m1.625s I'm still investigating this. Could be a simple coding error somewhere. Rich.

2 11

Weird results from g.sh()
by Sam Eiderman 06 Sep '20

06 Sep '20

Hi, I'm using libguestfs 1.42, When I run the following python3 commands on a rhel7.8: print(g.ls('/sys')) print(g.sh('ls /sys')) I get: [] block bus class dev devices firmware fs hypervisor kernel module power It seems that g.ls('/sys') is chrooted correctly. But g.sh('ls /sys') isn't. I came across this behavior when I used g.command(['grub2-mkconfig', '-o', '....']) on rhel7.8 and it didn't produce the correct linuxefi/initrdefi entries in grub, since in rhel7.8 the uefi detection algorithm is checking in bash '[ -d /sys/firmware/efi ]'. https://src.fedoraproject.org/rpms/grub2/blob/71e0fb4ea646071ec86e4a65bb242… I thought I could simply g.mkdir_p('/sys/firmware/efi') - but that created a fake dir in my chrooted environment, where grub2-mkconfig somehow was inspecting the real /sys as demonstrated in the above 'ls' example. Any ideas on why sh jailbreaks the chroot? is /sys set through some environment variable? Thanks!

2 5

[nbdkit PATCH 0/2] ext2 export list tweaks
by Eric Blake 01 Sep '20

01 Sep '20

Applies on top of my pending series for the exportname filter, addressing one of the todo's in that cover letter. Eric Blake (2): filters: Add .export_description wrappers ext2: Supply .list_exports and .default_export filters/ext2/nbdkit-ext2-filter.pod | 3 +- tests/Makefile.am | 16 +++- filters/ext2/ext2.c | 125 +++++++++++++++++++--------- filters/gzip/gzip.c | 40 ++++++--- filters/partition/partition.c | 58 +++++++++---- filters/tar/tar.c | 48 +++++++---- filters/xz/xz.c | 43 +++++++--- tests/test-ext2-exportname.sh | 73 ++++++++++++++++ 8 files changed, 305 insertions(+), 101 deletions(-) create mode 100755 tests/test-ext2-exportname.sh -- 2.28.0

2 5

[nbdkit PATCH v2 0/8] exportname filter
by Eric Blake 01 Sep '20

01 Sep '20

This is a revision of my .default_export work, plus new work on .export_descriptions and a new exportname filter. I think it is now ready to check in. Things I'd still like in 1.22: - the file plugin should implement .list_exports (patch already posted, but it needs rebasing on this series) - the ext2 filter should override .list_exports when in exportname mode - the nbd plugin should be able to use libnbd 1.4 features for .list_exports - more language bindings for new functions (perl, python, ocaml, ...) nice to have, but can be post-release: - the file plugin in dir= mode should have default=none|largest (default behavior none says "" is an error because it is not a file, setting it to largest picks the largest file in the directory for "") - the tar filter should consider an exportname mode - figure out how to let filters open the plugin without a current connection into the filter Eric Blake (8): api: Add .default_export api: Add nbdkit_add_default_export server: Respond to NBD_INFO_NAME request api: Add nbdkit_str[n]dup_intern helper sh, eval: Implement .default_export api: Alter .list_exports api: Add .export_description exportname: New filter docs/nbdkit-filter.pod | 56 ++- docs/nbdkit-plugin.pod | 123 ++++++- docs/nbdkit-protocol.pod | 14 + .../exportname/nbdkit-exportname-filter.pod | 154 ++++++++ filters/ext2/nbdkit-ext2-filter.pod | 5 + filters/log/nbdkit-log-filter.pod | 2 +- plugins/eval/nbdkit-eval-plugin.pod | 6 + plugins/file/nbdkit-file-plugin.pod | 9 +- plugins/sh/nbdkit-sh-plugin.pod | 39 +- include/nbdkit-common.h | 6 + include/nbdkit-filter.h | 13 +- include/nbdkit-plugin.h | 4 +- configure.ac | 2 + filters/exportname/Makefile.am | 67 ++++ tests/Makefile.am | 8 +- server/internal.h | 22 +- server/backend.c | 93 +++-- server/connections.c | 2 +- server/exports.c | 32 +- server/filters.c | 31 +- server/main.c | 26 +- server/nbdkit.syms | 3 + server/plugins.c | 44 ++- server/protocol-handshake-newstyle.c | 87 ++++- server/public.c | 48 +++ server/test-public.c | 9 +- plugins/sh/methods.h | 2 + plugins/cc/cc.c | 73 ++-- plugins/eval/eval.c | 70 ++-- plugins/ondemand/ondemand.c | 11 +- plugins/sh/methods.c | 84 ++++- plugins/sh/sh.c | 70 ++-- plugins/sh/example.sh | 2 +- filters/exportname/exportname.c | 342 ++++++++++++++++++ filters/log/log.c | 18 +- filters/tls-fallback/tls-fallback.c | 60 ++- tests/test-eval-exports.sh | 34 +- tests/test-export-info.sh | 111 ++++++ tests/test-exportname.sh | 184 ++++++++++ tests/test-tls-fallback.sh | 53 ++- tests/test-layers-filter.c | 13 +- tests/test-layers-plugin.c | 8 + tests/test-layers.c | 24 +- TODO | 20 +- 44 files changed, 1790 insertions(+), 294 deletions(-) create mode 100644 filters/exportname/nbdkit-exportname-filter.pod create mode 100644 filters/exportname/Makefile.am create mode 100644 filters/exportname/exportname.c create mode 100755 tests/test-export-info.sh create mode 100755 tests/test-exportname.sh -- 2.28.0

2 13

[nbdkit PATCH 0/3] .list_exports in nbd plugin
by Eric Blake 01 Sep '20

01 Sep '20

Another series on top of my exportname filter, marking off another todo bullet point. With this, you can now use the NBD plugin as a transparent passthrough of all export names served by the remote server in both directions (list advertisement server to client, and export name from client to server). Eric Blake (3): nbd: Implement .default_export, .export_description nbd: Add dynamic-export=true option nbd: Implement .list_exports plugins/nbd/nbdkit-nbd-plugin.pod | 30 +++- tests/Makefile.am | 4 + plugins/nbd/nbd.c | 250 +++++++++++++++++++++++++----- tests/test-nbd-dynamic-content.sh | 75 +++++++++ tests/test-nbd-dynamic-list.sh | 162 +++++++++++++++++++ 5 files changed, 482 insertions(+), 39 deletions(-) create mode 100755 tests/test-nbd-dynamic-content.sh create mode 100755 tests/test-nbd-dynamic-list.sh -- 2.28.0

2 4

nbdkit, Windows, Wine, scripts and tests
by Richard W.M. Jones 29 Aug '20

29 Aug '20

I spent a bit of my holiday trying to get some more tests working on the Windows port. Unfortunately I hit some unexpected and deep snags. I'm writing down these thoughts before I forget them. nbdkit runs shell scripts from a lot of places. The places which are important for tests are: (a) nbdkit --run 'script'. (b) linuxdisk plugin (used to create tests/disk, so important for tests) runs "du" and "mke2fs -d" commands. On Unix-like platforms these use system(3), and they also do in the Windows port since Win32 also has system(). When running on actual Windows the commands are run through CMD.EXE as you'd expect. Of course Windows doesn't have the rich and powerful shell environment that Unix has so there's no way for nbdkit --run to work exactly the same on Windows as it does on Unix, but that's all to be expected. More interesting for running the tests is what happens when running under Wine on Linux. I imagined that Linux -> Wine nbdkit.exe -> system ("qemu-img ...") would work by the Windows process running the Unix command directly. Unfortunately that's not how it works at all. In fact Wine emulates system() using CMD.EXE, and so none of the tests as written will work. It's possible to get system() under Wine to run Unix commands, but complicated to say the least: https://stackoverflow.com/a/45545068 However even if we get that lot to work, there's still a big catch. Paths are generally rewritten using nbdkit_realpath which on Windows returns paths like "C:\Windows\Temp\socket". Wheneven these appear in variables (eg. $unixsocket, $uri, and variables passed to the linuxdisk plugin subcommands) these will make no sense to the Unix commands we want to run. In summary, a test which does: nbdkit .. --run 'qemu-img info "$uri"' will currently try to run: CMD.EXE /C qemu-img info 'nbd+unix:///?socket=C:\Windows\Temp\sock' [Note that on actual Windows, assuming qemu-img.exe is available, I would expect this to work, although we still need to adjust how variables are set in the script fragment we pass to system(). The problem is not how to get this to work on Windows, but how to get it to work under Wine on Linux so we can run more tests.] Even if we resolve the problem with running shell commands, it will still not work because it will run: qemu-img info 'nbd+unix:///?socket=C:\Windows\Temp\sock' Rich. -- Richard Jones, Virtualization Group, Red Hat http://people.redhat.com/~rjones Read my programming and virtualization blog: http://rwmj.wordpress.com virt-top is 'top' for virtual machines. Tiny program with many powerful monitoring features, net stats, disk stats, logging, etc. http://people.redhat.com/~rjones/virt-top

1 0

Unknown libguestfs failure / race
by Sam Eiderman 27 Aug '20

27 Aug '20

Hi, I am getting a weird libguestfs failure from time to time: 2020-08-27 08:48:40 T libguestfs - 0 - appliance - SeaBIOS (version 1.12.0-1) 2020-08-27 08:48:40 T libguestfs - 0 - appliance - Booting from ROM... 2020-08-27 08:48:40 T libguestfs - 0 - library - child_cleanup: 0xd5a840: child process died Let me elaborate more about my environment I am running libguestfs 1.42.0 on GCP - that's nested virtualization. Using python bindings. Inside a docker. I wrote an output hook in python to emit logs more nicely. The supermin image is debian 10. The failure seems to happen from time to time just as I launch 16 "libguestfs"s on my 16 core machine. (This is a testing framework, so we are using libguestfs here to inspect and change different workloads, very similar to virt-v2v) Here is the full log of the sporadic failure: 2020-08-27 08:48:38 T libguestfs - 0 - enter - set_memsize 2020-08-27 08:48:38 T libguestfs - 0 - enter - add_drive 2020-08-27 08:48:38 T libguestfs - 0 - enter - launch 2020-08-27 08:48:38 T libguestfs - 0 - enter - max_disks 2020-08-27 08:48:38 T libguestfs - 0 - enter - get_tmpdir 2020-08-27 08:48:38 T libguestfs - 0 - enter - version 2020-08-27 08:48:38 T libguestfs - 0 - enter - get_backend 2020-08-27 08:48:38 T libguestfs - 0 - library - launch: program=my-v2v 2020-08-27 08:48:38 T libguestfs - 0 - library - launch: version=1.42.0 2020-08-27 08:48:38 T libguestfs - 0 - library - launch: backend registered: unix 2020-08-27 08:48:38 T libguestfs - 0 - library - launch: backend registered: uml 2020-08-27 08:48:38 T libguestfs - 0 - library - launch: backend registered: libvirt 2020-08-27 08:48:38 T libguestfs - 0 - library - launch: backend registered: direct 2020-08-27 08:48:38 T libguestfs - 0 - library - launch: backend=direct 2020-08-27 08:48:38 T libguestfs - 0 - library - launch: tmpdir=/tmp/libguestfszNoLuV 2020-08-27 08:48:38 T libguestfs - 0 - library - launch: umask=0022 2020-08-27 08:48:38 T libguestfs - 0 - library - launch: euid=0 2020-08-27 08:48:38 T libguestfs - 0 - library - begin testing qemu features 2020-08-27 08:48:38 T libguestfs - 0 - enter - get_cachedir 2020-08-27 08:48:38 T libguestfs - 0 - library - checking for previously cached test results of /usr/bin/kvm, in /var/tmp/.guestfs-0 2020-08-27 08:48:38 T libguestfs - 0 - library - loading previously cached test results 2020-08-27 08:48:38 T libguestfs - 0 - library - qemu version: 3.1 2020-08-27 08:48:38 T libguestfs - 0 - library - qemu mandatory locking: yes 2020-08-27 08:48:38 T libguestfs - 0 - library - qemu KVM: disabled 2020-08-27 08:48:38 T libguestfs - 0 - enter - get_backend_setting 2020-08-27 08:48:38 T libguestfs - 0 - enter - get_sockdir 2020-08-27 08:48:38 T libguestfs - 0 - library - finished testing qemu features 2020-08-27 08:48:38 T libguestfs - 0 - enter - get_backend_setting 2020-08-27 08:48:38 T libguestfs - 0 - library - command: run: dmesg | grep -Eoh 'lpj=[[:digit:]]+' 2020-08-27 08:48:38 T libguestfs - 0 - library - read_lpj_from_dmesg: calculated lpj=8800000 2020-08-27 08:48:38 T libguestfs - 0 - library - command: run: file 2020-08-27 08:48:38 T libguestfs - 0 - library - command: run: \ -- /appliance/root 2020-08-27 08:48:38 T libguestfs - 0 - appliance - /usr/bin/kvm \ 2020-08-27 08:48:38 T libguestfs - 0 - appliance - -global virtio-blk-pci.scsi=off \ 2020-08-27 08:48:38 T libguestfs - 0 - appliance - -no-user-config \ 2020-08-27 08:48:38 T libguestfs - 0 - appliance - -enable-fips \ 2020-08-27 08:48:38 T libguestfs - 0 - appliance - -nodefaults \ 2020-08-27 08:48:38 T libguestfs - 0 - appliance - -display none \ 2020-08-27 08:48:38 T libguestfs - 0 - appliance - -machine accel=kvm:tcg \ 2020-08-27 08:48:38 T libguestfs - 0 - appliance - -m 2048 \ 2020-08-27 08:48:38 T libguestfs - 0 - appliance - -no-reboot \ 2020-08-27 08:48:38 T libguestfs - 0 - appliance - -rtc driftfix=slew \ 2020-08-27 08:48:38 T libguestfs - 0 - appliance - -no-hpet \ 2020-08-27 08:48:38 T libguestfs - 0 - appliance - -global kvm-pit.lost_tick_policy=discard \ 2020-08-27 08:48:38 T libguestfs - 0 - appliance - -kernel /appliance/kernel \ 2020-08-27 08:48:38 T libguestfs - 0 - appliance - -initrd /appliance/initrd \ 2020-08-27 08:48:38 T libguestfs - 0 - appliance - -object rng-random,filename=/dev/urandom,id=rng0 \ 2020-08-27 08:48:38 T libguestfs - 0 - appliance - -device virtio-rng-pci,rng=rng0 \ 2020-08-27 08:48:38 T libguestfs - 0 - appliance - -device virtio-scsi-pci,id=scsi \ 2020-08-27 08:48:38 T libguestfs - 0 - appliance - -drive file=/tmp/xww-a1dtzg8i.qcow2,cache=writeback,format=qcow2,id=hd0,if=none \ 2020-08-27 08:48:38 T libguestfs - 0 - appliance - -device scsi-hd,drive=hd0 \ 2020-08-27 08:48:38 T libguestfs - 0 - appliance - -drive file=/appliance/root,snapshot=on,id=appliance,cache=unsafe,if=none,format=raw \ 2020-08-27 08:48:38 T libguestfs - 0 - appliance - -device scsi-hd,drive=appliance \ 2020-08-27 08:48:38 T libguestfs - 0 - appliance - -device virtio-serial-pci \ 2020-08-27 08:48:38 T libguestfs - 0 - appliance - -serial stdio \ 2020-08-27 08:48:38 T libguestfs - 0 - appliance - -device sga \ 2020-08-27 08:48:38 T libguestfs - 0 - appliance - -chardev socket,path=/tmp/libguestfsISN6Xx/guestfsd.sock,id=channel0 \ 2020-08-27 08:48:38 T libguestfs - 0 - appliance - -device virtserialport,chardev=channel0,name=org.libguestfs.channel.0 \ 2020-08-27 08:48:38 T libguestfs - 0 - appliance - -append "panic=1 console=ttyS0 edd=off udevtimeout=6000 udev.event-timeout=6000 no_timer_check lpj=8800000 printk.time=1 cgroup_disable=memory usbcore.nousb cryptomgr.notests tsc=reliable 8250.nr_uarts=1 root=UUID=dd6872eb-d621-499d-be42-64d78754c255 selinux=0 guestfs_verbose=1 TERM=linux vsyscall=emulate" 2020-08-27 08:48:40 T libguestfs - 0 - library - responding to serial console Device Status Report 2020-08-27 08:48:40 T libguestfs - 0 - appliance - ^[[1;256r^[[256;256H^[[6n 2020-08-27 08:48:40 T libguestfs - 0 - appliance - Google, Inc. 2020-08-27 08:48:40 T libguestfs - 0 - appliance - Serial Graphics Adapter 03/05/17 2020-08-27 08:48:40 T libguestfs - 0 - appliance - SGABIOS $Id: sgabios.S 8 2010-04-22 00:03:40Z nlaredo $ (generic@generic) Sun, 05 Mar 2017 16:09:17 +0100 2020-08-27 08:48:40 T libguestfs - 0 - appliance - Term: 80x24 2020-08-27 08:48:40 T libguestfs - 0 - appliance - 4 0 2020-08-27 08:48:40 T libguestfs - 0 - appliance - ^[[2J 2020-08-27 08:48:40 T libguestfs - 0 - appliance - SeaBIOS (version 1.12.0-1) 2020-08-27 08:48:40 T libguestfs - 0 - appliance - Booting from ROM... 2020-08-27 08:48:40 T libguestfs - 0 - library - child_cleanup: 0xd5a840: child process died 2020-08-27 08:48:40 T libguestfs - 0 - library - sending SIGTERM to process 459 2020-08-27 08:48:40 T libguestfs - 0 - library - qemu maxrss 60136K 2020-08-27 08:48:40 T libguestfs - 0 - subprocess_quit 2020-08-27 08:48:40 T libguestfs - 0 - library - closing guestfs handle 0xd5a840 (state 0) 2020-08-27 08:48:40 T libguestfs - 0 - close 2020-08-27 08:48:40 T libguestfs - 0 - library - command: run: rm 2020-08-27 08:48:40 T libguestfs - 0 - library - command: run: \ -rf /tmp/libguestfszNoLuV 2020-08-27 08:48:40 T libguestfs - 0 - library - command: run: rm 2020-08-27 08:48:40 T libguestfs - 0 - library - command: run: \ -rf /tmp/libguestfsISN6Xx Usually everything is working and the output is: 2020-08-27 07:41:10 T libguestfs - 0 - appliance - SeaBIOS (version 1.12.0-1) 2020-08-27 07:41:11 T libguestfs - 0 - appliance - Booting from ROM... 2020-08-27 07:41:12 T libguestfs - 0 - appliance - ^[[2J[ 0.000000] Linux version 4.19.0-10-amd64 (debian-kernel(a)lists.debian.org) (gcc version 8.3.0 (Debian 8.3.0-6)) #1 SMP Debian 4.19.132-1 (2020-07-24) 2020-08-27 07:41:12 T libguestfs - 0 - appliance - [ 0.000000] Command line: panic=1 console=ttyS0 edd=off udevtimeout=6000 udev.event-timeout=6000 no_timer_check lpj=8800000 printk.time=1 cgroup_disable=memory usbcore.nousb cryptomgr.notests tsc=reliable 8250.nr_uarts=1 root=UUID=dd6872eb-d621-499d-be42-64d78754c255 selinux=0 guestfs_verbose=1 TERM=linux vsyscall=emulate It seems that the qemu process has died somewhere between the ROM boot and the kernel loading. Any ideas how to debug this? Thanks! Sam

2 1

ANNOUNCE: nbdkit 1.22 - high performance NBD server
by Richard W.M. Jones 27 Aug '20

27 Aug '20

I'm pleased to announce the release of nbdkit 1.22, a high performance plugin-based Network Block Device (NBD) server. https://en.wikipedia.org/wiki/Network_block_device Key features of nbdkit: * Multithreaded NBD server written in C with good performance. * Minimal dependencies for the basic server. * Liberal license (BSD) allows nbdkit to be linked to proprietary libraries or included in proprietary code. * Well-documented, simple plugin API with a stable ABI guarantee. Lets you export “unconventional” block devices easily. * You can write plugins in C, Go, Lua, Perl, Python, OCaml, Ruby, Rust, shell script or Tcl. * Filters can be stacked in front of plugins to transform the output. Git: https://github.com/libguestfs/nbdkit Download: https://download.libguestfs.org/nbdkit/1.22-stable/ Fedora: https://koji.fedoraproject.org/koji/packageinfo?packageID=16469 Debian: https://packages.debian.org/nbdkit Release notes (http://libguestfs.org/nbdkit-release-notes-1.22.1.html) These are the release notes for nbdkit stable release 1.22. This describes the major changes since 1.20. nbdkit 1.22.0 was released on 27 August 2020. Security There were no security issues found. All past security issues and information about how to report new ones can be found in nbdkit-security(1). Plugins New nbdkit-cdi-plugin(1) (Containerized Data Importer) lets you export a single layer from a container image, allowing you to access disk images distributed using container registry infrastructure. New nbdkit-ondemand-plugin(1) creates filesystems for clients on demand. New nbdkit-torrent-plugin(1) serves a BitTorrent file or magnet link over NBD. nbdkit-data-plugin(1) and nbdkit-memory-plugin(1) have a new "allocator" parameter giving more fine-grained control over the allocation strategy. Currently implemented strategies are: sparse array, sparse array with zstd compression, malloc, and malloc with mlock. nbdkit-data-plugin(1) data format now supports strings and alignment. The "disk2data.pl" script has been enhanced to detect repeated sequences and produce a more compact output. nbdkit-curl-plugin(1) adds support for arbitrary custom request headers. Also it now supports scripts for generating cookies and custom request headers. This is especially useful when accessing web servers that require authentication which has to be renewed periodically. nbdkit-curl-plugin(1) now supports HTTP/2 servers (David Edmondson, Pino Toscano, Zi Liu). nbdkit-curl-plugin(1) adds support for the "User-Agent" header. nbdkit-file-plugin(1) can export multiple files in a directory as separate exports (Eric Blake). nbdkit-file-plugin(1) supports new "cache=none" and "fadvise=normal|random|sequential" parameters which can be used to optimize access patterns and avoid filling the page cache. nbdkit-nbd-plugin(1) supports proxying to another NBD server over "AF_VSOCK" (Eric Blake). nbdkit-nbd-plugin(1) supports new "command=" and "socket-fd=" parameters which allow you to run a subordinate NBD server as a command directly from the plugin, or pass in an already connected NBD endpoint. nbdkit-streaming-plugin(1) supports streaming to and from local pipes. nbdkit-vddk-plugin(1) now has support for VDDK 7.0. A new and useless "error" message printed by VDDK 7 has been suppressed (thanks Ming Xie, Pino Toscano). nbdkit-ext2-plugin(1) has been removed. It was deprecated in nbdkit 1.17.8. All existing uses can be replaced with nbdkit-ext2-filter(1). The "-" character is now permitted within plugin and filter names (Eric Blake). Filters New nbdkit-ddrescue-filter(1) lets you test disk images recovered by ddrescue(1) (François Revol). New nbdkit-gzip-filter(1) can decompress a plugin that contains gzip- compressed data. It replaces nbdkit-gzip-plugin(1) which is deprecated and will be removed in nbdkit 1.26. New nbdkit-pause-filter(1) can temporarily stop and later resume NBD requests. This can be used when taking snapshots. New nbdkit-swab-filter(1) swaps bytes in the underlying plugin, which can be used to adjust images that have an incorrect byte order (François Revol). New nbdkit-tar-filter(1) reads and writes files inside a tar file in the underlying plugin. It replaces nbdkit-tar-plugin(1) which is deprecated and will be removed in nbdkit 1.26. New nbdkit-tls-fallback-filter(1) provides alternate content when a client connects without using TLS to a server that permits but does not require TLS (Eric Blake). nbdkit-fua-filter(1) has new option "fuamode=pass" to pass through FUA and flush requests unchanged (disabling the filter). A new, unsafe "fuamode=discard" mode was added which discards FUA and flush requests. Language bindings New nbdkit-cc-plugin(3) lets you write small nbdkit plugins as inline C, C++ or OCaml "scripts" (and potentially other ahead-of-time compiled programming languages). nbdkit-rust-plugin(3) bindings have been replaced by higher level and more idiomatic bindings. These also support a greater range of server features (Alan Somers). nbdkit-python-plugin(3) now supports parallel plugins. These cannot run Python code in parallel (because of the Python GIL), but for Python code that blocks, eg. calling out to external services, requests can now be handled in parallel (thanks Nir Soffer). nbdkit-python-plugin(3) has a greater range of examples (Nir Soffer). nbdkit-python-plugin(3) adds support for getting the client export name, calling "nbdkit_debug", and also for extents. Server "nbdkit --dump-config" has new fields: "version_major" and "version_minor" to allow easier version parsing from scripts. "host_cpu" and "host_os" so you can find out if nbdkit was cross- compiled for another platform. "zstd=(yes|no)" so you can find out if nbdkit was compiled with zstd compression support. The -e/--exportname option (which never did anything useful previously) is now used to pass the preferred exportname to the --run option. In general you should never use this option except in very limited and exceptional circumstances (Eric Blake). API New ".after_fork" callback which can be used for plugins which need to create background threads, since those threads would otherwise be stranded if the server forks into the background. The VDDK plugin now defers VDDK library initialization til ".after_fork" because it is thought that the library creates background threads especially when accessing remote servers. New API "nbdkit_extents_aligned" is a helper function for filters that retrieves extents as needed until at least a certain range is covered (Eric Blake). New "nbdkit_is_tls" can be called to determine if TLS was negotiated on the connection. Plugins could use this to hide certain exports from non-authenticated/non-encrypted clients (Eric Blake). "nbdkit_read_password" with "password=-" will now return an error if the input is not a tty. Also "password=-FD" is disallowed if the file descriptor is 0, 1 or 2. Bug fixes The VDDK plugin was extensively refactored to fix many bugs: Reading passwords from a tty should work again. The plugin should now work properly when daemonized or using the --run option. The threading model has been relaxed to "SERIALIZE_REQUESTS" which should give small performance improvements when using multiple connections. The "base64" option of nbdkit-data-plugin(1) has been fixed so it sets the implicit size correctly. The "<FILE" formatter of nbdkit-data-plugin(1) now works if you use it more than once. Documentation New nbdkit-client(1) manual page documents how to mount filesystems from NBD exports on client machines, either directly or using systemd mountpoints. Examples now use "NBDKIT_API_VERSION 2" so that the examples match the documentation. Tests The tests should now pass on OpenBSD ≥ 6.4. tests/Makefile.am was refactored: There is now no longer a single massive "EXTRA_DIST". Creation of test disks was refactored. Some tests will now use new libnbd 1.4 tools nbdcopy(1) and nbdinfo(1) if available, and skip the tests otherwise. New test functions "requires_plugin" and "requires_filter" allows tests to be skipped if the corresponding plugin/filter was not compiled on a particular platform. So tests should now not fail when an optional external library needed by a plugin/filter is not available. Build Honggfuzz may now be used for fuzz testing. If nbdkit-python-plugin(3) is enabled, Python ≥ 3.6 must be used. Internals The export name is now passed down through filter ".open" callbacks, allowing filters to modify the export name passed through to plugins. The final thread model is passed to filter ".get_ready" callbacks. SEE ALSO nbdkit(1). AUTHORS Authors of nbdkit 1.22: Alan Somers (7 commits) Eric Blake (53 commits) François Revol (2 commits) Nir Soffer (2 commits) Richard W.M. Jones (285 commits) -- Richard Jones, Virtualization Group, Red Hat http://people.redhat.com/~rjones Read my programming and virtualization blog: http://rwmj.wordpress.com virt-top is 'top' for virtual machines. Tiny program with many powerful monitoring features, net stats, disk stats, logging, etc. http://people.redhat.com/~rjones/virt-top

1 0