scionproto / scion

SCION Internet Architecture
https://scion.org
Apache License 2.0
401 stars 161 forks source link

./scion.sh run failing #177

Closed pratyakshs closed 9 years ago

pratyakshs commented 9 years ago

The problem seems to be with beacon servers. Here's the output of logs/bs1-15-1.log (truncated): [WARNING] (BS shared pcbs) AD: Data too short for parsing, len: 20 [CRITICAL] (BS shared pcbs) Exception in handle_shared_pcbs thread: [CRITICAL] (BS shared pcbs) Traceback (most recent call last): [CRITICAL] (BS shared pcbs) File "/home/scion/lib/thread.py", line 51, in wrapper [CRITICAL] (BS shared pcbs) return f(*args, **kwargs) [CRITICAL] (BS shared pcbs) File "beacon_server.py", line 608, in handle_shared_pcbs [CRITICAL] (BS shared pcbs) count = self._read_cached_entries() [CRITICAL] (BS shared pcbs) File "beacon_server.py", line 635, in _read_cached_entries [CRITICAL] (BS shared pcbs) count = self._process_cached_pcbs(new, timed_desc=desc) [CRITICAL] (BS shared pcbs) File "/home/scion/lib/util.py", line 214, in wrapper [CRITICAL] (BS shared pcbs) ret = f(*args, **kwargs) [CRITICAL] (BS shared pcbs) File "beacon_server.py", line 664, in _process_cached_pcbs [CRITICAL] (BS shared pcbs) pcbs.append(PathSegment(raw=raw)) [CRITICAL] (BS shared pcbs) File "/home/scion/lib/packet/pcb.py", line 286, in __init__ [CRITICAL] (BS shared pcbs) self.parse(raw) [CRITICAL] (BS shared pcbs) File "/home/scion/lib/packet/pcb.py", line 309, in parse [CRITICAL] (BS shared pcbs) self.add_ad(ad_marking) [CRITICAL] (BS shared pcbs) File "/home/scion/lib/packet/pcb.py", line 327, in add_ad [CRITICAL] (BS shared pcbs) if ad_marking.pcbm.hof.exp_time < self.min_exp_time: [CRITICAL] (BS shared pcbs) AttributeError: 'NoneType' object has no attribute 'hof' [CRITICAL] (BS shared pcbs) [INFO] (MainThread) Received SIGTERM [INFO] (MainThread) Exiting

shitz commented 9 years ago

I think that one was missed by @pszalach when he cleaned up the various *Markings.

On Thu, Jun 18, 2015 at 11:41 AM, Pratyaksh Sharma <notifications@github.com

wrote:

The problem seems to be in bs1-15-1 Here's the output of logs/bs1-15-1.log (truncated): [WARNING](BS shared pcbs) AD: Data too short for parsing, len: 20 [CRITICAL](BS shared pcbs) Exception in handle_shared_pcbs thread: [CRITICAL](BS shared pcbs) Traceback (most recent call last): [CRITICAL](BS shared pcbs) File "/home/scion/lib/thread.py", line 51, in wrapper [CRITICAL](BS shared pcbs) return f(_args, _kwargs) [CRITICAL](BS shared pcbs) File "beacon_server.py", line 608, in handle_shared_pcbs [CRITICAL](BS shared pcbs) count = self._read_cached_entries() [CRITICAL](BS shared pcbs) File "beacon_server.py", line 635, in _read_cached_entries [CRITICAL](BS shared pcbs) count = self._process_cached_pcbs(new, timed_desc=desc) [CRITICAL](BS shared pcbs) File "/home/scion/lib/util.py", line 214, in wrapper [CRITICAL](BS shared pcbs) ret = f(_args, _kwargs) [CRITICAL](BS shared pcbs) File "beacon_server.py", line 664, in _process_cached_pcbs [CRITICAL](BS shared pcbs) pcbs.append(PathSegment(raw=raw)) [CRITICAL](BS shared pcbs) File "/home/scion/lib/packet/pcb.py", line 286, in init [CRITICAL](BS shared pcbs) self.parse(raw) [CRITICAL](BS shared pcbs) File "/home/scion/lib/packet/pcb.py", line 309, in parse [CRITICAL](BS shared pcbs) self.add_ad(ad_marking) [CRITICAL](BS shared pcbs) File "/home/scion/lib/packet/pcb.py", line 327, in add_ad [CRITICAL](BS shared pcbs) if ad_marking.pcbm.hof.exp_time < self.min_exp_time: [CRITICAL](BS shared pcbs) AttributeError: 'NoneType' object has no attribute 'hof' [CRITICAL](BS shared pcbs) INFO Received SIGTERM INFO Exiting

— Reply to this email directly or view it on GitHub https://github.com/netsec-ethz/scion/issues/177.

pszal commented 9 years ago

@pratyakshs this is in master? how to replicate that?

pszal commented 9 years ago

@shitz I didn't remove pcbm

pratyakshs commented 9 years ago

@pszalach yes, I pulled from master. And did ./scion.sh run after init, topology, setup.

pszal commented 9 years ago

@pratyakshs are you running with docker? I guess ZK's old cache is an issue.

pratyakshs commented 9 years ago

@pszalach no, not with docker. Can I clear ZK's cache?

pszal commented 9 years ago

stop ZK; sudo rm /var/lib/zookeeper/version-2/*; start ZK but it (path) may depend on your ZK instalation

pratyakshs commented 9 years ago

@pszalach It worked, thanks. Should we add sudo rm /var/lib/zookeeper/version-2/* to ./scion run?

kormat commented 9 years ago

No, definitely not. Having zookeeper maintain state is important for testing. The best approach is to use docker, as that implicitly gives you a fresh environment for running things.

pszal commented 9 years ago

Yes, I agree we shouldn't add that to run. Probably it would make much sense with ./scion.sh topology, but please use docker instead.

On 18.06.2015 12:07, Stephen Shirley wrote:

No, definitely not. Having zookeeper maintain state is important for testing. The best approach is to use docker, as that implicitly gives you a fresh environment for running things.

— Reply to this email directly or view it on GitHub https://github.com/netsec-ethz/scion/issues/177#issuecomment-113098504.