b64581b1 | 08-Jul-2015 |
Andreas Jaekel <jaekel@strato-rz.de> |
zev: zevadm now uses tmpqueue as a default for polling
zevadm poll mode used to create a non-persistent queue to receive events. The new tmpqueue device is now used instead. Functionality remains
zev: zevadm now uses tmpqueue as a default for polling
zevadm poll mode used to create a non-persistent queue to receive events. The new tmpqueue device is now used instead. Functionality remains the same. However, the tmpqueue will always be empty when created, while the old method allowed the option of seeing older zev events in the new queue, if available. This was rarely useful, though, and can still be done by creating an explicit persistent queue first.
(cherry picked from commit 22534ac1f0111cc26337b8b0247422f4f1ab1d7f)
show more ...
|
93bacbbd | 07-Jul-2015 |
Andreas Jaekel <jaekel@strato-rz.de> |
zev: zevadm show queue size as percentage
zevadm no longer displays the wakeup threshold. Instead, the length of a queue (in zevadm -l) is now displayed both as an absolute value and as a percentag
zev: zevadm show queue size as percentage
zevadm no longer displays the wakeup threshold. Instead, the length of a queue (in zevadm -l) is now displayed both as an absolute value and as a percentage of the maximum queue length.
(cherry picked from commit ff6c736fc4505b616c8d23a2b00f8bcd50449090)
show more ...
|
e3455c18 | 24-Jun-2015 |
Andreas Jaekel <jaekel@strato-rz.de> |
zev: new ioctl() to query ABI version
A new ioctl() has been added to query the module for it's version numbers. These are supposed to give an indication of which behaviour and which feature set ca
zev: new ioctl() to query ABI version
A new ioctl() has been added to query the module for it's version numbers. These are supposed to give an indication of which behaviour and which feature set can be expected to work.
(cherry picked from commit f63c8e0930c25065bb397bcbbe45afb4f170fc96)
show more ...
|
27daf72b | 02-Mar-2015 |
Jan Schlien <illumos.874@jan-o-sch.net> |
zevadm: add -E option for non-empty temp queues
In -p mode without a queue, a temporary queue is created to poll for current zev events. If one of the existing zev queues has a backlog, it may be de
zevadm: add -E option for non-empty temp queues
In -p mode without a queue, a temporary queue is created to poll for current zev events. If one of the existing zev queues has a backlog, it may be desired to peek at the oldes event still queued, which can be done with the combination of -p and -E.
(cherry picked from commit fb060709e9968994460ae1ec0f6e1ba3380575ab)
show more ...
|
34551a8b | 02-Mar-2015 |
Jan Schlien <illumos.874@jan-o-sch.net> |
zevadm: sort getopt letters to find a free one
(cherry picked from commit c2e7f0f3d40cfb4b0be69df9e2e829b18d88d8cd) |
35526fb3 | 30-Jan-2015 |
Arne Jansen <jansen@webgods.de> |
zevadm: add option to dump spool file
(cherry picked from commit f766c05e63f43019fadfd2c5bdb551ffc2cd0d0a) |
647b4f9e | 17-Dec-2014 |
Jan Schlien <illumos.874@jan-o-sch.net> |
zevadm: print payload for mark events in -p mode
In newer beaver versions, payload is the rid.
(cherry picked from commit 2b3e33c33d41d238dc28dcbe89719af7f60074ce) |
113f4c19 | 17-Dec-2014 |
Jan Schlien <illumos.874@jan-o-sch.net> |
zevadm: make temporary queue non-blocking and initially empty
When zevadm is used without -q, a temporarily created queue is used. The normal use case of this call is debugging current events. The s
zevadm: make temporary queue non-blocking and initially empty
When zevadm is used without -q, a temporarily created queue is used. The normal use case of this call is debugging current events. The stream of events should start after the call to zevadm -p, hence the shift to the initlally empty flag. Debugging should not lead to blocking systems, no matter how slow the events are printed to the reader's terminal, hence the non-blocking flag.
(cherry picked from commit a9097a45d68e3e619b07b2e7deaa22e3fce0deb2)
show more ...
|
8aa47a6b | 25-Nov-2014 |
Andreas Jaekel <jaekel@strato-rz.de> |
zev: added inode info of clobbered files (rename)
When a file is renamed, another file might previously have had the new name and will thus be clobbered. In this case, the inode information of the
zev: added inode info of clobbered files (rename)
When a file is renamed, another file might previously have had the new name and will thus be clobbered. In this case, the inode information of the clobbered file will be added to the zev rename event. Note that the file might still have other names.
(cherry picked from commit 9ed9bfe91df7eb8d10c125f9f04d3ec82010930a)
show more ...
|
a0d677d8 | 08-Sep-2014 |
Andreas Jaekel <jaekel@strato-rz.de> |
zevadm: flush stdout on sigint
(cherry picked from commit d0bd401f83b277c493d603ef0777bfdd04629cdb) |
94875cb8 | 08-Sep-2014 |
Andreas Jaekel <jaekel@strato-rz.de> |
zev: umount events include info of covered inode
A ZFS umount will now be reported including information of the inode that the filesystem was mounted over. (the mountpoint that will now become visib
zev: umount events include info of covered inode
A ZFS umount will now be reported including information of the inode that the filesystem was mounted over. (the mountpoint that will now become visible again)
(cherry picked from commit 808e6c04e8d170f4444fe287cf08403bdf1a757c)
show more ...
|
c99a1a25 | 05-Aug-2014 |
Andreas Jaekel <jaekel@strato-rz.de> |
zev: option to create empty queues
New zev queues where filled with as many events as would fit from the global event backlog. Now a flag can be set to skip this step. The new queue will then be em
zev: option to create empty queues
New zev queues where filled with as many events as would fit from the global event backlog. Now a flag can be set to skip this step. The new queue will then be empty upon creation and will only receive events newer than the queue itself.
(cherry picked from commit 6970b7d9a5856021f673d67e8ba9f061c20eac54)
show more ...
|
174dc952 | 05-Aug-2014 |
Andreas Jaekel <jaekel@strato-rz.de> |
zevadm: fixed segfault for missing parameter
(cherry picked from commit 9cced931c4e08136c492318b3126186124804e9a) |
f432e238 | 01-Aug-2014 |
Andreas Jaekel <jaekel@strato-rz.de> |
zevadm: add 'report' mode
zevadm has a new report mode which will write zev stats into files. The base filename is given, and will be appended with the current date. At midnight, a new file is cre
zevadm: add 'report' mode
zevadm has a new report mode which will write zev stats into files. The base filename is given, and will be appended with the current date. At midnight, a new file is created and used. The content is zevstat output for 60 second intervals.
(cherry picked from commit 079f408df9d389b9bdaa3a32b01df17fd58e802f)
show more ...
|
50da9ede | 29-Jul-2014 |
Andreas Jaekel <jaekel@strato-rz.de> |
zevadm: better option parsing for zevstat
when called as zevstat, zevadm now displays a zevstat usage. It also looks for a -v switch and the interval and count parameters actually work.
(cherry pic
zevadm: better option parsing for zevstat
when called as zevstat, zevadm now displays a zevstat usage. It also looks for a -v switch and the interval and count parameters actually work.
(cherry picked from commit 54da422bc6175dcf026cc1391a2526620a7dade1)
show more ...
|
19695134 | 29-Jul-2014 |
Andreas Jaekel <jaekel@strato-rz.de> |
zevadm: bugfixes for zevstat mode
Draining the new queue was done until poll() with a one second timeout return 0. On busy systems this never happens. Now poll() without a timeout must return 0.
W
zevadm: bugfixes for zevstat mode
Draining the new queue was done until poll() with a one second timeout return 0. On busy systems this never happens. Now poll() without a timeout must return 0.
When called as zevstat, giving no interval parameter caused a core dump.
(cherry picked from commit 9ccd4f509e10a1bd670ce35e6f03aae00d7ca2d4)
show more ...
|
b690436d | 29-Jul-2014 |
Andreas Jaekel <jaekel@strato-rz.de> |
zevadm: zevstat mode to show system activity
(cherry picked from commit ffacc4c18995665fffe3c01cdce2e14f424bc57c) |
a5090b97 | 22-Jul-2014 |
Andreas Jaekel <jaekel@strato-rz.de> |
zevadm: bugfix with option parsing
Option parsing was order dependent. For example, "-p -v" ignored the "-v" option while "-v -p" worked. Order does not matter anymore.
(cherry picked from commit
zevadm: bugfix with option parsing
Option parsing was order dependent. For example, "-p -v" ignored the "-v" option while "-v -p" worked. Order does not matter anymore.
(cherry picked from commit 4e01af0f92d75b7f57a617a8fbdd9dccc5077037)
show more ...
|
2eabeab5 | 22-Jul-2014 |
Andreas Jaekel <jaekel@strato-rz.de> |
zev: create and symlink events now inlude checksum
When a symlink or another new file is created, the resulting zev events now include a signature for the new file.
Symlinks are treated like files
zev: create and symlink events now inlude checksum
When a symlink or another new file is created, the resulting zev events now include a signature for the new file.
Symlinks are treated like files that contain the "link" part as text. Block and character devices have a signature based on the format string "%c%d,%d", with 'b'/'c', getmajor() and getminor() as values. All other file types will result in an all-zero signature.
Hardlinks create a new directory entry, but no new inode. Therefore, rdlink events still do not contain a signature.
(cherry picked from commit d27baf235074c236575a6c7ec7d3fe89a7d8af6e)
show more ...
|
12119a7e | 03-Jun-2014 |
Andreas Jaekel <jaekel@strato-rz.de> |
zev: add txg to applicable event type records
(cherry picked from commit be5f9b3c721e52d475c3d7009982992c01c9140e) |
16ff6b2f | 27-May-2014 |
Andreas Jaekel <jaekel@strato-rz.de> |
zev: zevadm -v now outputs all data in events
(cherry picked from commit 48ead878f6c5ed593c292e6cb4396521eca8767f) |
42110aac | 22-May-2014 |
Andreas Jaekel <jaekel@strato-rz.de> |
zev: ioctl() to get signatures for an open file
There's a new ioctl() ZEV_IOC_GET_FILE_SIGNATURES to get beaver signatures for a data range in an open file. The provided buffer has to be large enou
zev: ioctl() to get signatures for an open file
There's a new ioctl() ZEV_IOC_GET_FILE_SIGNATURES to get beaver signatures for a data range in an open file. The provided buffer has to be large enough to hold the parameters and the signatures.
zevadm can display a file's signatures using 'zevadm -c <filename>'
(cherry picked from commit 6b4c2eb9a413da85091774d0095ad61946ae043d)
show more ...
|
205ed6bf | 08-May-2014 |
Andreas Jaekel <jaekel@strato-rz.de> |
zev: calculate beaver-compatible checksums of changes
Ranges required to calculate the checksums will be range locked before the zev callbacks are called. Checksums are calculated in the zev module
zev: calculate beaver-compatible checksums of changes
Ranges required to calculate the checksums will be range locked before the zev callbacks are called. Checksums are calculated in the zev module and reported to userland. Checksums of all-zeroes blocks are omitted. Userland is expected to infer that the block must have bee all-zeroes.
An internal cache is implemented to avoid re-reading the same data over and over again.
(cherry picked from commit d8e62bab61474da73f26dc58b26d937c9cc17397)
show more ...
|
6a6a51ee | 08-May-2014 |
Andreas Jaekel <jaekel@strato-rz.de> |
zev: changes for #1891 review (multiplayer)
Review feedback addressed in this commit:
- remove avl, replace with array - read() can block (but can be interrupted by signals) - style changes - q
zev: changes for #1891 review (multiplayer)
Review feedback addressed in this commit:
- remove avl, replace with array - read() can block (but can be interrupted by signals) - style changes - queue threshold can be set to 0 bytes (no delay) - FIND_QUEUE macro replaced by function - new mutex to avoid races when queues are full - zevadm can be told to poll events from a specified queue - zevadm can create blocking queues
additional changes:
- mark events are delivered without delay
(cherry picked from commit e4ea145c5d454bda934f8af75eec9ea17f2489c0)
show more ...
|
add9520f | 19-Mar-2014 |
Andreas Jaekel <jaekel@strato-rz.de> |
zev: multi-queue support
(cherry picked from commit bc16f3206d5f60ea7a18a972c3a52afb85102b71) |