|
Another round 
The "Usage examples" page
This page several times has the term "VMWare". There should be "VMware" (lowercase "w").
The "Item types" -> "Zabbix agent" page
The "net.if.in" key has description for the <mode> parameter:
compressed - the number of compressed packets transmitted or received by the device driver
However, it is "net.if.in" key, it should show received statistics only; for transmitting statistics the "net.if.out" key does exist. Is it a documentation bug?
The "net.tcp.service" and "net.tcp.service.perf" (on the same page) have the broken links "known issues" in the comment column: the appropriate page does not have the "#https-checks" section.
The same page: key "net.tcp.port" has the following comment:
ip - IP or DNS name (default is 127.0.0.1)
However, the keys "net.tcp.service", "net.tcp.service.perf", "net.udp.service" and "net.udp.service.perf" does not have any mentions about DNS name; but in fact they support it. Probably, this should also be explicitly mentioned (together with the version number, when it becomes to be supported).
The same page, keys "net.tcp.service.perf" and "net.udp.service.perf": the type of returned value should be mentioned - Numeric (float).
The same page, key "net.tcp.socket.count" has comment with an example:
=> net.tcp.socket.count[,80,,,established] → check if local TCP port 80 is in "established" state
However, this metric returns a number of connections, not state of port. So, the more accurate description could be, for example:
returns the number of connections to local TCP port 80 in "established" state
The same page, key "net.udp.socket.count" has description:
Return the number of TCP sockets that match parameters.
However, number of TCP sockets is returned by the "net.tcp.socket.count" metric; the "net.udp.socket.count" metric should, probably, return the number of UDP sockets.
Also, the example for this key tells about the "listening" state; but the "Parameters" column tells that the "established" and "unconn" only permitted for the "<state>" parameter (and really "listening" for UDP is not supported).
The same page, a bit later: example for "sensor" metric for OpenBSD uses a regular expression for a device name; however, the parameters description does not contain any mentions about regular expressions in parameters.
The same page, a bit later: for the "system.hw.chassis" metric the possible parameters values should be emphasized by italics.
The same page, "system.localtime" metric has a broken link "dashboard widget" in the example; the correct link is: (/manual/web_interface/frontend_sections/monitoring/dashboard/widgets/clock).
The same page, "vfs.dir.count" and "vfs.dir.get" metrics have the following comment:
Both regex_incl and regex_excl are being applied to files and directories when calculating entry size, but are ignored when picking subdirectories to traverse
It seems as a result of copy-paste from the "vfs.dir.size" metric, because the "vfs.dir.count" metric does not return a size, it returns a count. The "count" should be instead of "size" in this phrase.
Something appropriate also should be for the "vfs.dir.get" metric.
The "vfs.dir.*" metrics have the following comment:
Execution time is limited by the default timeout value in agent configuration (3 sec).
Is it really true? I.e. limit is always exactly "the default timeout value in agent configuration (3 sec)" and could not be changed if the the parameter "Timeout=" is explicitly set by to a longer value?
The "vfs.dir.size" has the following example:
This formulation is not very accurate:
- it will count files with 'log' in their names (not in file content);
- this example will not exclude all "files containing '.old'"; it will exclude files with names endings by '.old'.
The "vfs.file.exists" has the following comment:
The file size limit depends on large file support.
However, this metric does not depend on the file size. Does it really matter?
The same is true for the "vfs.file.time" metric.
The "vfs.fs.inode" metric has bad description of "mode" parameter:
- the "//" unnecessary;
- "pfree" should be emphasized by italics.
Probably, this comment should be exactly as for the "vfs.fs.size" metric.
The "vfs.fs.size" metric has the following comment:
In case of a mounted volume, disk space for local file system is returned.
It is not clear what does it mean. Probably, just opposite: if the specified filesystem is NOT mounted, then size of local filesystem (i.e. where the mount point is located) is returned?
Or it's assumed a Windows platform?
The same page, "Web monitoring data"" section: the unclear term "Punycode" is mentioned several times.
It will be great to supply it by the link to Wikipedia (https://en.wikipedia.org/wiki/Punycode).
The same page, footnote 3 has misspelled device name "sbd" (there should be "sdb").
| 
Team A