vergoh / vnstat Goto Github PK

Vnstat has made my day. I configured it just right, started the daemon, and waited patiently for data to amass in each interface's database.

I then noticed graph columns did not align correctly:

I've encountered this problem in my own programs. As programmers, we cannot assume that characters of localized strings only occupy a single column. The fix is straightforward and begins with the man page for wcswidth().

Thank you for maintaining such an awesome utility.

Hi,
I got this problem 4 times on this month during connecting via 3G USB (only occur on 3G USB). The counter increase 8Gb suddenly with Download 4GB and Upload 4GB.
I don't know why and facing it few time in this month. How can I find the cause of this problem?

Excuse me if this is possible but it's not obvious in any manual pages or docs that I've come across. Would it be possible to be able to log and report on periods longer than 12 months, like say over a 10 year period?

I have set my router to reset wlan0 every 6 hours using cron and wifi command.
Sometimes after running that command vnstat is adding 4GB of transfer to my wlan0 statistic. Anyone know why?

It's happening on OpenWRT Chaos Calmer and was present on OpenWRT Barrier Breaker.

Example picture attached. March is called März in german and (I guess) the umlaut leads to some miscalculation. The locale I'm using is de_DE.UTF-8.

vnstat -s

                  rx      /      tx      /     total    /   estimated

wan0:
Jan '16 4.00 GiB / 3.40 GiB / 7.40 GiB
Mar '16 0.99 GiB / 1.05 GiB / 2.04 GiB / 2.26 GiB
yesterday 18.52 MiB / 23.78 MiB / 42.30 MiB
today 2.90 MiB / 2.39 MiB / 5.29 MiB / 32 MiB

however in standard printout:

vnstat
Database updated: Tue Mar 1 02:55:01 2016

wan0 since 08/03/15

      rx:  16.17 GiB      tx:  11.53 GiB      total:  27.70 GiB

monthly
rx | tx | total | avg. rate
------------------------+-------------+-------------+---------------
Jan '16 4.00 GiB | 3.40 GiB | 7.40 GiB | 23.18 kbit/s
Feb '16 0.99 GiB | 1.05 GiB | 2.04 GiB | 7.57 kbit/s
------------------------+-------------+-------------+---------------
estimated 1.10 GiB | 1.16 GiB | 2.26 GiB |

daily
rx | tx | total | avg. rate
------------------------+-------------+-------------+---------------
yesterday 18.52 MiB | 23.78 MiB | 42.30 MiB | 4.01 kbit/s
today 2.90 MiB | 2.39 MiB | 5.29 MiB | 4.13 kbit/s
------------------------+-------------+-------------+---------------
estimated 16 MiB | 16 MiB | 32 MiB |

looks like February in -s short printout is missing

We get an error when parsing vnstat -i br0 -h --json using python -m json.tool.
Indeed, the output contains some wrongly formatted json on traffic.hours array which contains a leading , (..."hours":[,{"id":15,"...)

{"vnstatversion":"1.13","jsonversion":"1","interfaces":[{"id":"br0","nick":"br0","created":{"date":{"year":2015,"month":2,"day":8}},"updated":{"date":{"year":2015,"month":2,"day":8},"time":{"hour":15,"minutes":21}},"traffic":{"total":{"rx":974,"tx":128},"days":[{"id":0,"date":{"year":2015,"month":2,"day":8},"rx":974,"tx":128}],"months":[{"id":0,"date":{"year":2015,"month":2},"rx":974,"tx":128}],"tops":[],"hours":[,{"id":15,"date":{"year":2015,"month":2,"day":8},"rx":974,"tx":128}]}}]}

Not all required includes are included.

See openwrt/packages#1415

hello，i use vnstat to monitor my server , but something that confuse me .

and this happen in 2017.06.24 12:00 am , ,so much traffic appear in an hour , but traffic appear normal on my server provider.

server infomation :
centos 6.8

Hi,

according to the docs, vnstatd will only create database files for interfaces when there aren't any files yet. How do get it to automatically create database files for interfaces that are added later on? Is there an option for that?

I find myself want 5, or 15 minute intervals as well (coincides with better loadavg for one).

Unfortunately 'm' is already used in the db for month.

hello,
the devices are not in sort order shown.
i have no idea if there is any option for it ?

My router with LEDE (version 17.01.2) has vnstat version 1.12. I could upgrade it to vnstat version 1.17 by following instructions on https://forum.lede-project.org/t/why-vnstat-is-at-version-1-12-1-only/1179/21

My Linux Mint 18.2 Xfce has vnstat version 1.14. I could follow the instructions on https://askubuntu.com/questions/938293/update-vnstat-to-1-17 to upgrade to vnstat 1.17.

Here's my question. If I upgrade the vnstat on my Linux Mint computer or my router, will I lose some vnstat data/logs/stats that are being saved?

Another question: Before I can upgrade to vnstat 1.17, wouldn't I first have to remove my current vnstat? If so, am I right to think that uninstalling would remove all the vnstat logs/stats/database?

When vnstat shows me stats it gives wrong information about the avg. rate.

Example:

         day         rx      |     tx      |    total    |   avg. rate
     ------------------------+-------------+-------------+---------------
      01/17/16      1.77 GiB |    1.09 GiB |    2.86 GiB |  277.92 kbit/s

That is wrong information because 2.86 GiB (Gebibyte) per day equals 284.34 kbit/s.

Either your calculation is wrong or you're just using the wrong prefix. I suspect the latter, because 277.92 kibibyte/s would be right.

I attached a screenshot of my calculations, just to show what I did.

Using JSON as output makes vnstat one of the best network monitoring tools.
However I've noticed that --json parameter is not working in some cases like:

1. live traffic monitoring ( vnstat -l --json )
2. Calculate traffic ( vnstat -tr --json )

Maybe there is some workaround to achieve JSON output in such situations, if not It might turnout to be very handy to add such support.

Is there any way to change the "first day of the month" for vnstat? Example:

$ vnstat -m

 external (enp4s0)  /  monthly

       month        rx      |     tx      |    total    |   avg. rate
    ------------------------+-------------+-------------+---------------
      May '16     45.71 GiB |    1.86 GiB |   47.57 GiB |  148.99 kbit/s
      Jun '16    261.80 GiB |    4.57 GiB |  266.36 GiB |    1.26 Mbit/s
    ------------------------+-------------+-------------+---------------
    estimated    381.14 GiB |    6.64 GiB |  387.79 GiB |

As you can see, vnstat is treating the first day I started the daemon as the first day of the month. I need it to be the actual first day, e.g. for June it should start at June 1.

My /etc/vnstat.conf has MonthRotate 1 set, but that doesn't seem to have any effect.

$ vnstat -v
vnStat 1.15 by Teemu Toivola <tst at iki dot fi>

If so, could you offer a tag in github for new releases from now on?

I also would like to see tarball available on your homepage.

Thanks.

... either that, or the documentation is incorrect.

I am just getting into vnstat and think I am going to love it. In setting up the initial configuration, I noticed that PollInterval was capped at 60 instead of 300 as listed in the documentation.

Thank you!

I just installed vnstat and enable the service. I wan't to reset it to zero like I never used it.

vnstat -r >> doesn't reset.

vnstat --delete ; vnstat --create -i mydevice ; reboot >> doesn't reset either.
After --create. the profile that was created was the profile that was deleted before.

rm /var/lib/vnstat/mydevice ; vnstat --create -i mydevice ; reboot also doesn't reset.
same thing happened after --create.
just like this guy: http://www.linuxquestions.org/questions/linux-networking-3/vnstat-doesn%27t-reset-494165/

What could be the problem? Am I doing it wrong?
Thank you :'D

Install docs are now incorrect since you changed to autotools because the binaries are no longer in src/ but in the root of the working directory.

I can no longer build with image output support on FreeBSD since you changed to autotools. libgd is installed, we're passing the correct LDFLAGS as always and it simply won't play ball.

Is there a reason you changed to autotools? Your old makefiles worked perfectly fine.

https://www.varnish-cache.org/docs/2.1/phk/autocrap.html

This happens for every installation and the fix would be

chown vnstat: /var/lib/vnstat/*

OS: CentOS 6 & 7
Installed using yum.

And insert the JSON output in one unnecessary commas, therefore in jq command cannot parse error occurs.

This is in recognition that the bug right?

# vnstat --json
{"vnstatversion":"1.13","jsonversion":"1","interfaces":[{"id":"eth0","nick":"eth0","created":{"date":{"year":2015,"month":4,"day":6}},"updated":{"date":{"year":2015,"month":4,"day":6},"time":{"hour":16,"minutes":2}},"traffic":{"total":{"rx":33464,"tx":920},"days":[{"id":0,"date":{"year":2015,"month":4,"day":6},"rx":33464,"tx":920}],"months":[{"id":0,"date":{"year":2015,"month":4},"rx":33464,"tx":920}],"tops":[],"hours":[,{"id":15,"date":{"year":2015,"month":4,"day":6},"rx":33367,"tx":894},{"id":16,"date":{"year":2015,"month":4,"day":6},"rx":97,"tx":26}]}}]}
[root@localhost ~]# vnstat --json | jq .
parse error: Expected value before ',' at line 1, column 425

Greetings,

I have a feature request -
It would be very helpful to my platform and usage needs if it were possible to get the -tr argument output in JSON format.

Thank you for work on this great product.

nevermind, sorry!

In addition with the default monthly monitoring, I would like to monitor a custom date range matching the billing period. How can it be done?

If not, please let me know how to configure monitoring the billing period.

When I run

vnstat -d -s

it returns

                      rx      /      tx      /     total    /   estimated
 eth0:
       Dec '15     19.47 MiB  /  543.02 MiB  /  562.49 MiB  /   11.01 GiB
         today     19.47 MiB  /  543.02 MiB  /  562.49 MiB  /    1.01 GiB

as you can see, it adds an extra 10 to the days estimate.

I noticed traffic from apt-mirror is not (reliably) counted by vnstat, but the traffic shows up in my Zabbix monitoring. apt-mirror downloaded a good 120GB at a speed of roughly 700Mbit/s today, yet only 18GB were registered by vnstat (which includes other traffic by other applications). I haven't noticed any other applications not being counted on my system. I have no idea how this is even possible, vnstat usually even counts routed traffic on the network interface. Any suggestions?

I updated from 1.12 to 1.16 and I got this error.
Error: wlan0: Invalid database timestamp.
Error: validation of imported database failed.

How do I fix this? I can't even import my backup databases.

I couldn't find a mention of this in the changelog; and the config-file didn't seem to have an option for this either. This change makes the output pretty useless as it's just a bunch of numbers now (which are also being truncated). Can I change this; or is this a bug?

Im using the command --exportdb --xml, my xml is then converted to json.

The first id in traffic.days.day and traffic.months.month always seem to be current month/day. However if i do the same to traffic.months.month the usage reported will be in the id of the current hours. (So say i download a huge file at 1700 o'clock the reported use will be in id 16)

Output https://gist.github.com/Hellowlol/7808a07c190a911cc208

Is this intended?

I see errors like this

In file included from common.h:18:0,
                 from misc.c:2:
/usr/include/sys/file.h:209:2: error: unknown type name 'u_int'
  u_int xf_flag; /* flags (see fcntl.h) */

in gmake all, even with the CPATH and LIBRARY_PATH variables set. I'm not a C programmer, so what other information do you need?

I am using the --json parameter to output results for my website's dashboard. Using a script that converts the output into a readable format, it returns MB, rather than GB. The readable numbers are the exact same.

My question is, is the rx and tx numbers bytes?

Just fresh installed and -> Failed to restart vnstat.service: Unit vnstat.service is masked.

ps aux|grep vnstat
--nothing there ...

I was wondering if there is any chance that support for per-process stats would be added, because it would be quite useful to have.

When using a grsecurity enhanced kernel and running vnstatd with the user vnstat, no data is collected.
Due to proc restrictions by grsec, the vnstat user must be a member of the grsec-proc group.

When this is not the case, vnstatd collects no data and produces no error or warning log messages.
Please add one for this case.

Maybe at

Hey
Is there any API of sorts which is possible to use to integrate in a different application. Or some way to access its output stats?

Thanks.

Hey,
I'm attempting to statically cross compile vnstat-1.16 (same with 1.15) to ARM or MIPS64, using musl-1.15 but the build fails. The cross-compiler works fine as I was able to compile many other softwares. It is using gcc-6.2.0.
In case you want to look what was done, here the output → https://framagit.org/Ypnose/solyste/snippets/363/raw

i have a question.
i use a shell script to get the output from one month from different servers.
one server has more b/w usage then the other.
one server gives me 700 mb usage per month
another server gives me 4 gb usage per month.

in my output i get 700 from one server and 4 from the other.
is it possible to get the output in mb ? so i get 700 from one server and 4096 from the other server ?

here is a part of the line i use.
vnstat -m | grep "date +"%b '%y"" | awk '{print $3}' > monitor/download.txt

maybe you have another solution that i can get the output in bytes so i can calculate in the script to get back to mb or gb.

hello,
thanks for this great software !
i'm using vnStat 1.12 on debian jessie and didn't see anything about what i'm looking for:
i ve got several logical interface, and i need details per sub-interface
eth0
eth0:1
eth0:2

is that possible ?

Hi @vergoh,

Thanks for this useful tool, I was wondering if -h can show data as -d, -m to use a human readable unit? Currently, it shows the data in KiB, hope it can auto transfer to use MiB and GiB.

Thanks!

[ifinfo.c:131]: (error) Dangerous usage of 'ifaceid' (strncpy doesn't always null-terminate it).

Source code is

strncpy(ifaceid, iface, 32);
strcat(ifaceid, ":");

Suggest new code

strncpy(ifaceid, iface, 32);
ifaceid[32] = '\0';
strcat(ifaceid, ":");

Can you consider adding vnstat to Coverity Scan.

In the function readproc there is a buffer overflow:
char temp[4][64], procline[512], *proclineptr, ifaceid[33];
[...]
while (fgets(procline, 512, fp)!=NULL) {
sscanf(procline, "%511s", temp[0]);

The variable temp is only 4*64 (=256) bytes, yet the code tries to fill it with 512 byte and thus overflows. (To see the overflow at runtime you can compile vnstat with address sanitizer: 'make CFLAGS="-fsanitize=address" LDFLAGS="-fsanitize=address"' and then run 'src/vnstat -tr')

I think a patch like this should fix it (but not 100% sure, cause I don't know what exactly the code is meant to do):

--- ./vnstat-1.14/src/ifinfo.c  2015-03-29 20:56:55.000000000 +0200
+++ ./vnstat-1.14-1/src/ifinfo.c    2015-12-07 18:35:50.981122234 +0100
@@ -182,8 +182,8 @@
    strcat(ifaceid, ":");

    check = 0;
-   while (fgets(procline, 512, fp)!=NULL) {
-       sscanf(procline, "%511s", temp[0]);
+   while (fgets(procline, 256, fp)!=NULL) {
+       sscanf(procline, "%255s", temp[0]);
        if (strncmp(ifaceid, temp[0], strlen(ifaceid))==0) {
            /* if (debug)
                printf("\n%s\n", procline); */

Please delete

I noticed that vnstat has duplicated entries for some days and months. Look at the pictures I attached.
I run Ubuntu 14.04 with the 3.13.0-61 kernel. The vnstat version is 1.14.

I've been doing some weird stuff and it's messed with my vnstat output. I realized this when I had already uploaded three gigabytes abnormally. Can I edit the database to return the value for a specific hour to a sensible size? I don't know what format databases are in, or how to open them.

Would be nice to have remote controll of vnstat so you can collect data from diffrent hosts!

Such as in FreeNAS jails or other simular setups.

Sorry bad english!

Since Gigabit links are common, what about changing the default MaxBandwidth setting from 100 to 1000? As far as I can see, this limit exists to detect abnormally high deviations and there shouldn't be any other negative effects from doing so?