1

The -q argument of traceroute sets the number of probes to be used.

On Linux systems it seems that this can only be up to 10. If a higher value is used traceroute returns.

no more than 10 probes per hop

On a solaris box the same parameter is available and there it accepts more than 10.

I wonder why that limit is there. Is that just arbitrary choice?

user55570
  • 793
  • So the question that needs answering: do you ever need more than 10? – Rinzwind Jun 10 '15 at 06:44
  • I have specifically encountered a scenario on Solaris boxes where UDP probes from traceroute are consistently being lost or dropped at the 11th probe. I was therefore keen to investigate whether this behaviour would continue if I used a box with a different OS. It now seems I cannot pursue it using a Linux box--else I would have to mess with code and try to recompile traceroute. – user55570 Jun 10 '15 at 10:43

2 Answers2

0

You can not expect that any program on any system generates identical outputs. Everyone is free to adapt the source code.

POSIX defines the application programming interface (API), along with command line shells and utility interfaces, for software compatibility with variants of Unix and other operating systems.

A.B.
  • 90,397
0

It's simply defined in the source code of the tool, among other limits:

#define MAX_HOPS    255
#define MAX_PROBES  10
#define MAX_GATEWAYS_4  8
#define MAX_GATEWAYS_6  127
#define DEF_HOPS    30
#define DEF_SIM_PROBES  16  /*  including several hops   */
#define DEF_NUM_PROBES  3
#define DEF_WAIT_SECS   5.0
#define DEF_SEND_SECS   0
#define DEF_DATA_LEN    40  /*  all but IP header...  */
#define MAX_PACKET_LEN  65000

There isn't a comment indicating why, so I guess, "arbitrary choice" is it.

Probably the developers thought that more than 10 per hop would not be useful and would probably not represent a sane choice.

Are you dealing with high packet loss or jitter? Even 10 packets should be more than enough to establish whether there is high packet loss or jitter.

thomasrutter
  • 36,774
  • I have specifically encountered a scenario on Solaris boxes where UDP probes from traceroute are being lost or dropped at the 11th probe. I was therefore keen to investigate whether this behaviour would continue if I used a box with a different OS. It now seems I cannot pursue it using a Linux box--else I would have to mess with code and try to recompile traceroute. – user55570 Jun 10 '15 at 10:38