[jira] Commented: (NUTCH-558) Need tool to retrieve domain statistics
[ https://issues.apache.org/jira/browse/NUTCH-558?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#action_12530755 ] Chris Schneider commented on NUTCH-558: --- The reason that DomainStats does not use URLUtils is that (as mentioned above) we are currently using a relatively old Nutch source base (last integrated at revision 417928). There are probably other tools/resources we could use as well if we reworked the code to better fit the current Nutch/Hadooop source environment. Sorry for being so out of date. Need tool to retrieve domain statistics --- Key: NUTCH-558 URL: https://issues.apache.org/jira/browse/NUTCH-558 Project: Nutch Issue Type: New Feature Affects Versions: 0.9.0 Reporter: Chris Schneider Assignee: Chris Schneider Attachments: DomainStats.patch Several developers have expressed interest in a tool to retrieve statistics from a crawl on a domain basis (e.g., how many pages were successfully fetched from www.apache.org vs. apache.org, where the latter total would include the former). -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Commented: (NUTCH-558) Need tool to retrieve domain statistics
[ https://issues.apache.org/jira/browse/NUTCH-558?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#action_12529749 ] Chris Schneider commented on NUTCH-558: --- I made a comment in the source about this, but thinking about it later, I do wonder whether this version truly works correctly when presented with a segment directory (in addition to a crawldb). I had to rewrite the InputFormat section of the tool to fit the latest Nutch/Hadoop source environment, and in the process, I removed the wrapper object necessary for my older source environment. I'd certainly welcome it if somebody out there with a more up to date installation and crawl data could give it a try. Need tool to retrieve domain statistics --- Key: NUTCH-558 URL: https://issues.apache.org/jira/browse/NUTCH-558 Project: Nutch Issue Type: New Feature Affects Versions: 0.9.0 Reporter: Chris Schneider Assignee: Chris Schneider Attachments: DomainStats.patch Several developers have expressed interest in a tool to retrieve statistics from a crawl on a domain basis (e.g., how many pages were successfully fetched from www.apache.org vs. apache.org, where the latter total would include the former). -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Created: (NUTCH-558) Need tool to retrieve domain statistics
Need tool to retrieve domain statistics --- Key: NUTCH-558 URL: https://issues.apache.org/jira/browse/NUTCH-558 Project: Nutch Issue Type: New Feature Affects Versions: 0.9.0 Reporter: Chris Schneider Assignee: Chris Schneider Several developers have expressed interest in a tool to retrieve statistics from a crawl on a domain basis (e.g., how many pages were successfully fetched from www.apache.org vs. apache.org, where the latter total would include the former). -- This message is automatically generated by JIRA. - You can reply to this email to add a comment to the issue online.
[jira] Commented: (NUTCH-351) Protocol forward proxy
[ http://issues.apache.org/jira/browse/NUTCH-351?page=comments#action_12446424 ] Chris Schneider commented on NUTCH-351: --- I just noticed a bug in the patch above. I believe it's missing a return sequence between the Host: host and Accept-Encoding: x-gzip, gzip These lines: reqStr.append( HTTP/1.0\r\n); reqStr.append(Host: ); reqStr.append(host); reqStr.append(portString); reqStr.append(Accept-Encoding: x-gzip, gzip\r\n); Need to look something like: reqStr.append( HTTP/1.0\r\n); reqStr.append(Host: ); reqStr.append(host); reqStr.append(portString); reqStr.append(\r\n); reqStr.append(Accept-Encoding: x-gzip, gzip\r\n); Protocol forward proxy -- Key: NUTCH-351 URL: http://issues.apache.org/jira/browse/NUTCH-351 Project: Nutch Issue Type: New Feature Components: fetcher Affects Versions: 0.8, 0.9.0, 0.8.1 Reporter: Sami Siren Assigned To: Sami Siren Priority: Minor Fix For: 0.9.0 Attachments: protocol-http-proxy-adapter.txt Protocol proxy adapter takes advantage of protocols known to http forward proxy. Usually there's atleast http, https and ftp. You must configure nutch to use this plugin and to use http proxy before use. -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: http://issues.apache.org/jira/secure/Administrators.jspa - For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] Created: (NUTCH-385) Server delay feature conflicts with maxThreadsPerHost
Server delay feature conflicts with maxThreadsPerHost - Key: NUTCH-385 URL: http://issues.apache.org/jira/browse/NUTCH-385 Project: Nutch Issue Type: Bug Components: fetcher Reporter: Chris Schneider For some time I've been puzzled by the interaction between two paramters that control how often the fetcher can access a particular host: 1) The server delay, which comes back from the remote server during our processing of the robots.txt file, and which can be limited by fetcher.max.crawl.delay. 2) The fetcher.threads.per.host value, particularly when this is greater than the default of 1. According to my (limited) understanding of the code in HttpBase.java: Suppose that fetcher.threads.per.host is 2, and that (by chance) the fetcher ends up keeping either 1 or 2 fetcher threads pointing at a particular host continuously. In other words, it never tries to point 3 at the host, and it always points a second thread at the host before the first thread finishes accessing it. Since HttpBase.unblockAddr never gets called with (((Integer)THREADS_PER_HOST_COUNT.get(host)).intValue() == 1), it never puts System.currentTimeMillis() + crawlDelay into BLOCKED_ADDR_TO_TIME for the host. Thus, the server delay will never be used at all. The fetcher will be continuously retrieving pages from the host, often with 2 fetchers accessing the host simultaneously. Suppose instead that the fetcher finally does allow the last thread to complete before it gets around to pointing another thread at the target host. When the last fetcher thread calls HttpBase.unblockAddr, it will now put System.currentTimeMillis() + crawlDelay into BLOCKED_ADDR_TO_TIME for the host. This, in turn, will prevent any threads from accessing this host until the delay is complete, even though zero threads are currently accessing the host. I see this behavior as inconsistent. More importantly, the current implementation certainly doesn't seem to answer my original question about appropriate definitions for what appear to be conflicting parameters. In a nutshell, how could we possibly honor the server delay if we allow more than one fetcher thread to simultaneously access the host? It would be one thing if whenever (fetcher.threads.per.host 1), this trumped the server delay, causing the latter to be ignored completely. That is certainly not the case in the current implementation, as it will wait for server delay whenever the number of threads accessing a given host drops to zero. -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: http://issues.apache.org/jira/secure/Administrators.jspa - For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] Commented: (NUTCH-385) Server delay feature conflicts with maxThreadsPerHost
[ http://issues.apache.org/jira/browse/NUTCH-385?page=comments#action_12441528 ] Chris Schneider commented on NUTCH-385: --- This comment was actually made by Andrzej in response to an email containing the analysis above that I sent him before creating this JIRA issue: Let's start with defining what is the desired semantics of these two parameters together. In my opinion it's the following: * if only 1 thread per host is allowed, at any given moment at most one thread should be accessing the host, and the interval between consecutive requests should be at least crawlDelay (whichever way we determine this value - from config, from robots.txt or external sources such as partner agreements). * if two or more (for example N) threads per host are allowed, at any given moment at most N threads should be accessing the host, and the interval between consecutive requests should be at least crawlDelay - that is, the interval between when one of the threads finishes, and another starts requesting. I.e.: for threads.per.host=2 and crawlDelay=3 seconds, if we start 3 threads trying to access the same host we should get something like this (time in [s] on the x axis, # - start request, + - request in progress, b - blocked in per-host limit, c - obeying crawlDelay): ===0 1 2 ===01234567890123456789012345678 1: #+++cccbbccc#cccbb#++ 2: #cccbcccbcc#+++cb 3: ccc#+ccc#+ccc#+++ As you can see, at any given time we have at most 2 threads accessing the site, and the interval between consecutive requests is at least 3 seconds. Especially interesting in the above graph is the period between 17-18 seconds - thread 2 had to be delayed additional 2 seconds to satisfy the crawl delay requirement, even though the threads.per.host requirement was satisfied. [snip] It's a question of priorities - in the model I drafted above the topmost priority is the observance of crawlDelay, sometimes at the cost of the number of concurrent threads (see seconds 17-18). In this model, the code should always put the delay in BLOCKED_ADDR_TO_TIME, in order to wait at least crawlDelay after _any_ thread finishes. We could use an alternative model, where crawlDelay is measured from the start of the request, and not from the end - see the graph below: ===0 1 2 3 ===01234567890123456789012345678901234567 1: #+++cbbb##++cc#+++ 2: ccc#cc#+++c#c# 3: cc#+ccc#+ccc#+ccbb but it seems to me that it's more complicated, gives less requests/sec, and the interpretaion of crawlDelay's meaning is stretched ... [snip] Server delay feature conflicts with maxThreadsPerHost - Key: NUTCH-385 URL: http://issues.apache.org/jira/browse/NUTCH-385 Project: Nutch Issue Type: Bug Components: fetcher Reporter: Chris Schneider For some time I've been puzzled by the interaction between two paramters that control how often the fetcher can access a particular host: 1) The server delay, which comes back from the remote server during our processing of the robots.txt file, and which can be limited by fetcher.max.crawl.delay. 2) The fetcher.threads.per.host value, particularly when this is greater than the default of 1. According to my (limited) understanding of the code in HttpBase.java: Suppose that fetcher.threads.per.host is 2, and that (by chance) the fetcher ends up keeping either 1 or 2 fetcher threads pointing at a particular host continuously. In other words, it never tries to point 3 at the host, and it always points a second thread at the host before the first thread finishes accessing it. Since HttpBase.unblockAddr never gets called with (((Integer)THREADS_PER_HOST_COUNT.get(host)).intValue() == 1), it never puts System.currentTimeMillis() + crawlDelay into BLOCKED_ADDR_TO_TIME for the host. Thus, the server delay will never be used at all. The fetcher will be continuously retrieving pages from the host, often with 2 fetchers accessing the host simultaneously. Suppose instead that the fetcher finally does allow the last thread to complete before it gets around to pointing another thread at the target host. When the last fetcher thread calls HttpBase.unblockAddr, it will now put System.currentTimeMillis() + crawlDelay into BLOCKED_ADDR_TO_TIME for the host. This, in turn, will prevent any threads from accessing this host until the delay is complete, even though zero threads are currently accessing the host. I see this behavior as inconsistent. More importantly, the current implementation certainly doesn't seem to answer my original question about appropriate definitions for what appear to be conflicting parameters. In a nutshell, how could we possibly honor the server delay
[jira] Commented: (NUTCH-385) Server delay feature conflicts with maxThreadsPerHost
[ http://issues.apache.org/jira/browse/NUTCH-385?page=comments#action_12441529 ] Chris Schneider commented on NUTCH-385: --- This comment was actually made by Ken Krugler, who was responding to Andrzej's comment above: [with respect to Andrzej's definitions at the beginning of his comment - Ed.:] I agree that this is one of two possible interpretations. The other is that there are N virtual users, and there crawlDelay applies to each of these virtual users in isolation. Using the same type of request data from above, I see a queue of requests with the following durations (in seconds): 4, 9, 6, 5, 6, 4, 7, 4 So with the virtual user model (where N = 2, thus A and B users), I get: ===0 1 2 ===01234567890123456789012345678 A: 4+++ccc6+ccc6+ccc7++ B: 9ccc5ccc4+++ccc4+++ The numbers mark the start of each new request, and the total duration for the request. This would seem to be less efficient than your approach, but somehow feels more in the nature of what threads.per.host really means. Let's see, for N = 3 this would look like: ===0 1 2 ===01234567890123456789012345678 A: 4+++ccc5ccc7++ccc B: 9ccc4+++ccc C: 6+ccc6+ccc4+++ccc [snip] To implement the virtual users model, each unique domain being actively fetched from would need to have N bits of state tracking the time of completion of the last request. Anyway, just an alternative interpretation... Server delay feature conflicts with maxThreadsPerHost - Key: NUTCH-385 URL: http://issues.apache.org/jira/browse/NUTCH-385 Project: Nutch Issue Type: Bug Components: fetcher Reporter: Chris Schneider For some time I've been puzzled by the interaction between two paramters that control how often the fetcher can access a particular host: 1) The server delay, which comes back from the remote server during our processing of the robots.txt file, and which can be limited by fetcher.max.crawl.delay. 2) The fetcher.threads.per.host value, particularly when this is greater than the default of 1. According to my (limited) understanding of the code in HttpBase.java: Suppose that fetcher.threads.per.host is 2, and that (by chance) the fetcher ends up keeping either 1 or 2 fetcher threads pointing at a particular host continuously. In other words, it never tries to point 3 at the host, and it always points a second thread at the host before the first thread finishes accessing it. Since HttpBase.unblockAddr never gets called with (((Integer)THREADS_PER_HOST_COUNT.get(host)).intValue() == 1), it never puts System.currentTimeMillis() + crawlDelay into BLOCKED_ADDR_TO_TIME for the host. Thus, the server delay will never be used at all. The fetcher will be continuously retrieving pages from the host, often with 2 fetchers accessing the host simultaneously. Suppose instead that the fetcher finally does allow the last thread to complete before it gets around to pointing another thread at the target host. When the last fetcher thread calls HttpBase.unblockAddr, it will now put System.currentTimeMillis() + crawlDelay into BLOCKED_ADDR_TO_TIME for the host. This, in turn, will prevent any threads from accessing this host until the delay is complete, even though zero threads are currently accessing the host. I see this behavior as inconsistent. More importantly, the current implementation certainly doesn't seem to answer my original question about appropriate definitions for what appear to be conflicting parameters. In a nutshell, how could we possibly honor the server delay if we allow more than one fetcher thread to simultaneously access the host? It would be one thing if whenever (fetcher.threads.per.host 1), this trumped the server delay, causing the latter to be ignored completely. That is certainly not the case in the current implementation, as it will wait for server delay whenever the number of threads accessing a given host drops to zero. -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: http://issues.apache.org/jira/secure/Administrators.jspa - For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] Commented: (NUTCH-351) Protocol forward proxy
[ http://issues.apache.org/jira/browse/NUTCH-351?page=comments#action_12438002 ] Chris Schneider commented on NUTCH-351: --- I would really appreciate it if Sami could explain in a little more detail what this patch adds to the proxy support already in Nutch. Although the patch seems to generalize the support somewhat, my reading of the current HttpResponse.java code suggests that it is already designed to handle URLs using these protocols when Nutch lives behind a proxy server. Protocol forward proxy -- Key: NUTCH-351 URL: http://issues.apache.org/jira/browse/NUTCH-351 Project: Nutch Issue Type: New Feature Components: fetcher Affects Versions: 0.8, 0.8.1, 0.9.0 Reporter: Sami Siren Assigned To: Sami Siren Priority: Minor Fix For: 0.9.0 Attachments: protocol-http-proxy-adapter.txt Protocol proxy adapter takes advantage of protocols known to http forward proxy. Usually there's atleast http, https and ftp. You must configure nutch to use this plugin and to use http proxy before use. -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: http://issues.apache.org/jira/secure/Administrators.jspa - For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] Updated: (NUTCH-371) DeleteDuplicates should remove documents with duplicate URLs
[ http://issues.apache.org/jira/browse/NUTCH-371?page=all ] Chris Schneider updated NUTCH-371: -- Description: DeleteDuplicates is supposed to delete documents with duplicate URLs (after deleting documents with identical MD5 hashes), but this part is apparently not yet implemented. Here's the comment from DeleteDuplicates.java: // 2. map indexes - url, fetchdate, index,doc // partition by url // reduce, deleting all but most recent. // // Part 2 is not yet implemented, but the Indexer currently only indexes one // URL per page, so this is not a critical problem. It is apparently also known that re-fetching the same URL (e.g., one month later) will result in more than one document with the same URL (this is alluded to in NUTCH-95), but the comment above suggests that the indexer will solve the problem before DeleteDuplicates, because it will only index one document per URL. This is not necessarily the case if the segments are to be divided among search servers, as each server will have its own index built from its own portion of the segments. Thus, if the URL in question was fetched in different segments, and these segments end up assigned to different search servers, then the indexer can't be relied on to eliminate the duplicates. Thus, it seems like the second part of the DeleteDuplicates algorithm (i.e., deleting documents with duplicate URLs) needs to be implemented. I agree with Byron and Andrzej that the most recently fetched document (rather than the one with the highest score) should be preserved. Finally, it's also possible to get duplicate URLs in the segments without re-fetching an expired URL in the crawldb. This can happen if 3 different URLs all redirect to the target URL. This is yet another consequence of handling redirections immediately, rather than adding the target URL to the crawldb for fetching in some subsequent segment (see NUTCH-273). was: DeleteDuplicates is supposed to delete documents with duplicate URLs (after deleting documents with identical MD5 hashes), but this part is apparently not yet implemented. Here's the comment from DeleteDuplicates.java: // 2. map indexes - url, fetchdate, index,doc // partition by url // reduce, deleting all but most recent. // // Part 2 is not yet implemented, but the Indexer currently only indexes one // URL per page, so this is not a critical problem. It is apparently also known that re-fetching the same URL (e.g., one month later) will result in more than one document with the same URL (this is alluded to in NUTCH-95), but the comment above suggests that the indexer will solve the problem before DeleteDuplicates, because it will only index one document per URL. This is not necessarily the case if the segments are to be divided among search servers, as each server will have its own index built from its own portion of the segments. Thus, if the URL in question was fetched in different segments, and these segments end up assigned to different search servers, then the indexer can't be relied on to eliminate the duplicates. Thus, it seems like the second part of the DeleteDuplicates algorithm (i.e., deleting documents with duplicate URLs) needs to be implemented. I agree with Byron and Andrzej that most recently fetched document (rather than the one with the highest score) should be preserved. Finally, it's also possible to get duplicate URLs in the segments without re-fetching an expired URL in the crawldb. This can happen if 3 different URLs all redirect to the target URL. This is yet another consequence of handling redirections immediately, rather than adding the target URL to the crawldb for fetching some subsequent segment (see NUTCH-273). DeleteDuplicates should remove documents with duplicate URLs Key: NUTCH-371 URL: http://issues.apache.org/jira/browse/NUTCH-371 Project: Nutch Issue Type: Bug Components: indexer Reporter: Chris Schneider DeleteDuplicates is supposed to delete documents with duplicate URLs (after deleting documents with identical MD5 hashes), but this part is apparently not yet implemented. Here's the comment from DeleteDuplicates.java: // 2. map indexes - url, fetchdate, index,doc // partition by url // reduce, deleting all but most recent. // // Part 2 is not yet implemented, but the Indexer currently only indexes one // URL per page, so this is not a critical problem. It is apparently also known that re-fetching the same URL (e.g., one month later) will result in more than one document with the same URL (this is alluded to in NUTCH-95), but the comment above suggests that the indexer will solve the problem before DeleteDuplicates, because it will only index one document per URL. This is not necessarily the case if the segments are to be divided among search servers,
[jira] Commented: (NUTCH-273) When a page is redirected, the original url is NOT updated.
[ http://issues.apache.org/jira/browse/NUTCH-273?page=comments#action_12430117 ] Chris Schneider commented on NUTCH-273: --- Another reason why it would be better to wait until the next segment to process the target of the redirect is that this target may already have been fetched. In this case, there's no need to refetch it. More importantly, though, refetching the page will cause its OPIC score to be distributed a second time to its outlinks. In fact, each page that redirects to the target page will cause the target page's OPIC score to get redistributed. I honestly can't see a good reason for doing an immediate redirect, since hopefully these cases aren't common enough to make a significant difference to crawling performance. Note that there are several other issues related to this issue, so we should take care to satisfy the goals of all with any fix. In particular, I agree that we should be saving more information in the metadata about the redirection (as well as other protocol cases). When a page is redirected, the original url is NOT updated. --- Key: NUTCH-273 URL: http://issues.apache.org/jira/browse/NUTCH-273 Project: Nutch Issue Type: Bug Components: fetcher Affects Versions: 0.8 Environment: n/a Reporter: Lukas Vlcek [Excerpt from maillist, sender: Andrzej Bialecki] When a page is redirected, the original url is NOT updated - so, CrawlDB will never know that a redirect occured, it won't even know that a fetch occured... This looks like a bug. In 0.7 this was recorded in the segment, and then it would affect the Page status during updatedb. It should do so 0.8, too... -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: http://issues.apache.org/jira/secure/Administrators.jspa - For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] Created: (NUTCH-348) Generator is building fetch list using *lowest* scoring URLs
Generator is building fetch list using *lowest* scoring URLs Key: NUTCH-348 URL: http://issues.apache.org/jira/browse/NUTCH-348 Project: Nutch Issue Type: Bug Components: fetcher Reporter: Chris Schneider Ever since revision 391271, when the CrawlDatum key was replaced by a FloatWritable key, the Generator.Selector.reduce method has been outputting the *lowest* scoring URLs! The CrawlDatum class has a Comparator that essentially treats higher scoring CrawlDatum objects as less than lower scoring CrawlDatum objects, so the higher scoring ones would appear first in a sequence file sorted using this as the key. When a FloatWritable based on the score itself (as returned from scfilters.generatorSortValue) became the sort key, it should have been negated in Generator.Selector.map to have the same result. Curiously, there is a comment to this effect immediately before the FloatWritable is set: // sort by decreasing score sortValue.set(sort); It seems like the simplest way to fix this is to just negate the score, and this seems to work for me: // sort by decreasing score // 2006-08-15 CSc REALLY sort by decreasing score sortValue.set(-sort); Unfortunately, this means that any crawls that have been done using Generator.java after revision 391271 should be discarded, as they were focused on fetching the lowest scoring unfetched URLs in the crawldb, essentially pointing the crawler 180 degrees from its intended direction. -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: http://issues.apache.org/jira/secure/Administrators.jspa - For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] Commented: (NUTCH-342) Nutch commands log to nutch/logs/hadoop.logs by default
[ http://issues.apache.org/jira/browse/NUTCH-342?page=comments#action_12426039 ] Chris Schneider commented on NUTCH-342: --- I apologize for my confusion. I had been thinking that hadoop-env.sh was getting sourced when a Nutch command was run; it is not. Thus, $HADOOP_LOG_DIR and $HADOOP_LOG_FILE are not set when executing Nutch commands. For now, I think it makes most sense for me to set NUTCH_LOG_DIR and NUTCH_LOGFILE to the same locations as $HADOOP_LOG_DIR and $HADOOP_LOG_FILE via .bash_profile, etc. I consider this awkward, but am unsure about how best to address this design problem. I'm beginning to think that NUTCH_LOGFILE should default to something like nutch-$USER-$COMMAND-`hostname`.log, which would seem more appropriate to find within the $NUTCH_HOME/logs directory. Nutch commands log to nutch/logs/hadoop.logs by default --- Key: NUTCH-342 URL: http://issues.apache.org/jira/browse/NUTCH-342 Project: Nutch Issue Type: Bug Affects Versions: 0.8 Reporter: Chris Schneider Priority: Minor Attachments: NUTCH-342.patch If (by default) Nutch commands are going to send their output to a file named hadoop.log, then it seems like the default location for this file should be the same location where Hadoop is putting its hadoop.log file (i.e., $HADOOP_LOG_DIR). Currently, if I set HADOOP_LOG_DIR to a special location (via hadoop-env.sh), this has no effect on where Nutch commands send their output. Some would probably suggest that I could just set NUTCH_LOG_DIR to $HADOOP_LOG_DIR myself. I still think that it should be defaulted this way in the nutch script. However, I'm unaware of an elegant way to modify such Nutch environment variables anyway. The hadoop-env.sh file provides a convenient place to modify Hadoop environment variables, but doing the same for Nutch environment variables presumably requires you to modify .bash_profile or a similar user script file (which is the way I used to accomplish this kind of thing with Nutch 0.7). -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: http://issues.apache.org/jira/secure/Administrators.jspa - For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] Created: (NUTCH-342) Nutch commands log to nutch/logs/hadoop.logs by default
Nutch commands log to nutch/logs/hadoop.logs by default --- Key: NUTCH-342 URL: http://issues.apache.org/jira/browse/NUTCH-342 Project: Nutch Issue Type: Bug Affects Versions: 0.8 Reporter: Chris Schneider Priority: Minor If (by default) Nutch commands are going to send their output to a file named hadoop.log, then it seems like the default location for this file should be the same location where Hadoop is putting its hadoop.log file (i.e., $HADOOP_LOG_DIR). Currently, if I set HADOOP_LOG_DIR to a special location (via hadoop-env.sh), this has no effect on where Nutch commands send their output. Some would probably suggest that I could just set NUTCH_LOG_DIR to $HADOOP_LOG_DIR myself. I still think that it should be defaulted this way in the nutch script. However, I'm unaware of an elegant way to modify such Nutch environment variables anyway. The hadoop-env.sh file provides a convenient place to modify Hadoop environment variables, but doing the same for Nutch environment variables presumably requires you to modify .bash_profile or a similar user script file (which is the way I used to accomplish this kind of thing with Nutch 0.7). -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: http://issues.apache.org/jira/secure/Administrators.jspa - For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] Updated: (NUTCH-342) Nutch commands log to nutch/logs/hadoop.logs by default
[ http://issues.apache.org/jira/browse/NUTCH-342?page=all ] Chris Schneider updated NUTCH-342: -- Attachment: NUTCH-342.patch Here's a patch that defaults NUTCH_LOG_DIR to $HADOOP_LOG_DIR and NUTCH_LOGFILE to $HADOOP_LOG_FILE. Nutch commands log to nutch/logs/hadoop.logs by default --- Key: NUTCH-342 URL: http://issues.apache.org/jira/browse/NUTCH-342 Project: Nutch Issue Type: Bug Affects Versions: 0.8 Reporter: Chris Schneider Priority: Minor Attachments: NUTCH-342.patch If (by default) Nutch commands are going to send their output to a file named hadoop.log, then it seems like the default location for this file should be the same location where Hadoop is putting its hadoop.log file (i.e., $HADOOP_LOG_DIR). Currently, if I set HADOOP_LOG_DIR to a special location (via hadoop-env.sh), this has no effect on where Nutch commands send their output. Some would probably suggest that I could just set NUTCH_LOG_DIR to $HADOOP_LOG_DIR myself. I still think that it should be defaulted this way in the nutch script. However, I'm unaware of an elegant way to modify such Nutch environment variables anyway. The hadoop-env.sh file provides a convenient place to modify Hadoop environment variables, but doing the same for Nutch environment variables presumably requires you to modify .bash_profile or a similar user script file (which is the way I used to accomplish this kind of thing with Nutch 0.7). -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: http://issues.apache.org/jira/secure/Administrators.jspa - For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] Updated: (NUTCH-336) Harvested links shouldn't get db.score.injected in addition to inbound contributions
[ http://issues.apache.org/jira/browse/NUTCH-336?page=all ] Chris Schneider updated NUTCH-336: -- Attachment: NUTCH-336.patch.txt Here's a patch that fixes the problem. It separates a new injectionScore API out from the initialScore API. Harvested links shouldn't get db.score.injected in addition to inbound contributions Key: NUTCH-336 URL: http://issues.apache.org/jira/browse/NUTCH-336 Project: Nutch Issue Type: Bug Components: fetcher Affects Versions: 0.8 Reporter: Chris Schneider Priority: Minor Attachments: NUTCH-336.patch.txt Currently (even with Stefan's fix for NUTCH-324), harvested links have their initial scores set to db.score.injected + (sum of inbound contributions * db.score.link.[internal | external]), but this will place (at least external) harvested links even higher than injected URLs on the fetch list. Perhaps more importantly, this effect cascades. As a simple example, if each page in A-B-C-D has exactly one external link and only A is injected, then D will receive an initial score of at least (4*db.score.injected) with the default db.score.link.external of 1.0. Higher values of db.score.injected and db.score.link.external obviously exacerbate this problem. -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: http://issues.apache.org/jira/secure/Administrators.jspa - For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] Created: (NUTCH-336) Harvested links shouldn't get db.score.injected in addition to inbound contributions
Harvested links shouldn't get db.score.injected in addition to inbound contributions Key: NUTCH-336 URL: http://issues.apache.org/jira/browse/NUTCH-336 Project: Nutch Issue Type: Bug Components: fetcher Affects Versions: 0.8 Reporter: Chris Schneider Priority: Minor Currently (even with Stefan's fix for NUTCH-324), harvested links have their initial scores set to db.score.injected + (sum of inbound contributions * db.score.link.[internal | external]), but this will place (at least external) harvested links even higher than injected URLs on the fetch list. Perhaps more importantly, this effect cascades. As a simple example, if each page in A-B-C-D has exactly one external link and only A is injected, then D will receive an initial score of at least (4*db.score.injected) with the default db.score.link.external of 1.0. Higher values of db.score.injected and db.score.link.external obviously exacerbate this problem. -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: http://issues.apache.org/jira/secure/Administrators.jspa - For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] Created: (NUTCH-301) CommonGrams loads analysis.common.terms.file for each query
CommonGrams loads analysis.common.terms.file for each query --- Key: NUTCH-301 URL: http://issues.apache.org/jira/browse/NUTCH-301 Project: Nutch Type: Improvement Components: searcher Versions: 0.8-dev Reporter: Chris Schneider The move away from static objects toward instance variables has resulted in CommonGrams constructor parsing its analysis.common.terms.file for each query. I'm not certain how large a performance impact this really is, but it seems like something you'd want to avoid doing for each query. Perhaps the solution is to keep around an instance of the CommonGrams object itself? -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: http://issues.apache.org/jira/secure/Administrators.jspa - For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] Created: (NUTCH-267) Indexer doesn't consider linkdb when calculating boost value
Indexer doesn't consider linkdb when calculating boost value Key: NUTCH-267 URL: http://issues.apache.org/jira/browse/NUTCH-267 Project: Nutch Type: Bug Components: indexer Versions: 0.8-dev Reporter: Chris Schneider Priority: Minor Before OPIC was implemented (Nutch 0.7, very early Nutch 0.8-dev), if indexer.boost.by.link.count was true, the indexer boost value was scaled based on the log of the # of inbound links: if (boostByLinkCount) res *= (float)Math.log(Math.E + linkCount); This is no longer true (even before Andrzej implemented scoring filters). Instead, the boost value is just the square root (or some other scorePower) of the page score. Shouldn't the invertlinks command, which creates the linkdb, have some affect on the boost value calculated during indexing (either via the OPICScoringFilter or some other built-in filter)? -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: http://issues.apache.org/jira/secure/Administrators.jspa - For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] Commented: (NUTCH-246) segment size is never as big as topN or crawlDB size in a distributed deployement
[ http://issues.apache.org/jira/browse/NUTCH-246?page=comments#action_12374253 ] Chris Schneider commented on NUTCH-246: --- As it turns out, this problem was due to a time synchronization between the jobtracker and the tasktrackers. When the URLs were injected, their fetchTimes were set to the System.currentTime() of the tasktrackers, which were 2 minutes in the future. Soon afterward, during the generation phase, these fetchTimes were compared to curTime, which came from the (correct) clock on the jobtracker (via the crawl.gen.curTime property in job.xml?) Thus, if the injection proceeded quickly enough, the generation phase would begin before these URLs were ready to be fetched. It seems like the Injector should be loading the current time from a job configuration property in the same way that that the Generator is doing now, then calling setFetchTime(), rather than leaving this to what the CrawlDatum constructor sets it to. segment size is never as big as topN or crawlDB size in a distributed deployement - Key: NUTCH-246 URL: http://issues.apache.org/jira/browse/NUTCH-246 Project: Nutch Type: Bug Versions: 0.8-dev Reporter: Stefan Groschupf Priority: Blocker Fix For: 0.8-dev I didn't reopen NUTCH-136 since it is may related to the hadoop split. I tested this on two different deployement (with 10 ttrackers + 1 jobtracker and 9 ttracks and 1 jobtracker). Defining map and reduce task number in a mapred-default.xml does not solve the problem. (is in nutch/conf on all boxes) We verified that it is not a problem of maximum urls per hosts and also not a problem of the url filter. Looks like the first job of the Generator (Selector) already got to less entries to process. May be this is somehow releasted to split generation or configuration inside the distributed jobtracker since it runs in a different jvm as the jobclient. However we was not able to find the source for this problem. I think that should be fixed before publishing a nutch 0.8. -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: http://issues.apache.org/jira/secure/Administrators.jspa - For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] Updated: (NUTCH-246) segment size is never as big as topN or crawlDB size in a distributed deployement
[ http://issues.apache.org/jira/browse/NUTCH-246?page=all ] Chris Schneider updated NUTCH-246: -- Priority: Minor (was: Blocker) segment size is never as big as topN or crawlDB size in a distributed deployement - Key: NUTCH-246 URL: http://issues.apache.org/jira/browse/NUTCH-246 Project: Nutch Type: Bug Versions: 0.8-dev Reporter: Stefan Groschupf Priority: Minor Fix For: 0.8-dev I didn't reopen NUTCH-136 since it is may related to the hadoop split. I tested this on two different deployement (with 10 ttrackers + 1 jobtracker and 9 ttracks and 1 jobtracker). Defining map and reduce task number in a mapred-default.xml does not solve the problem. (is in nutch/conf on all boxes) We verified that it is not a problem of maximum urls per hosts and also not a problem of the url filter. Looks like the first job of the Generator (Selector) already got to less entries to process. May be this is somehow releasted to split generation or configuration inside the distributed jobtracker since it runs in a different jvm as the jobclient. However we was not able to find the source for this problem. I think that should be fixed before publishing a nutch 0.8. -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: http://issues.apache.org/jira/secure/Administrators.jspa - For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] Commented: (NUTCH-246) segment size is never as big as topN or crawlDB size in a distributed deployement
[ http://issues.apache.org/jira/browse/NUTCH-246?page=comments#action_12374049 ] Chris Schneider commented on NUTCH-246: --- A few more details: Stefan and I were able to reproduce this problem using either an injection set of 4500 URLs or a larger set of DMOZ URLs. With the 4500 URL injection, only 653 URLs were generated for the first segment, despite the fact that topN was set to 500K. I confirmed that nearly all of the 4500 injected URLs passed our URL filer and were actually injected into the crawldb. To eliminate the possibility that the bug had been fixed recently or was due to a code modification that we'd made ourselves, we deployed yesterday's sandbox version of nutch (2006-04-10), including hadoop-0.1.1.jar (though I believe that Stefan had to build it himself because the nutch-0.8-dev.jar didn't match the source). We made the absolute minimum changes to nutch-site.xml, hadoop-site.xml, and hadoop-env.sh in order to deploy this version properly in our cluster (1 jobtracker/namenode machine, 10 tasktracker/datanode machines). However, we got the same results (i.e., very few URLs actually generated). This bug has apparently been present since at least change 382948, but I suspect that it may have been present for the entire history of the mapreduce implementation of Nutch. It may also be the root cause of NUTCH-136, the explanation for which has always left me somewhat dissatisfied. Just because a nutch-site.xml containing default properties may override the desired mapred properties (incorrectly) specified in one of the *-default.xml files, and may therefore set mapred.map.tasks and mapred.reduce.tasks back to the defaults (2 and 1, respectively), it's not clear to me exactly how/why you'd get only a fraction of topN URLs fetched. As Stefan has suggested, it would actually seem more plausible if each tasktracker tried to fetch the entire set of URLs in this case. I would suggest that someone with a good understanding of the hadoop implementation investigate the first generation job in fine detail, both for the case where the mapred properties are specified in an appropriate manner and for the case where nutch-site.xml overrides the desired properties, setting them back to the defaults. segment size is never as big as topN or crawlDB size in a distributed deployement - Key: NUTCH-246 URL: http://issues.apache.org/jira/browse/NUTCH-246 Project: Nutch Type: Bug Versions: 0.8-dev Reporter: Stefan Groschupf Priority: Blocker Fix For: 0.8-dev I didn't reopen NUTCH-136 since it is may related to the hadoop split. I tested this on two different deployement (with 10 ttrackers + 1 jobtracker and 9 ttracks and 1 jobtracker). Defining map and reduce task number in a mapred-default.xml does not solve the problem. (is in nutch/conf on all boxes) We verified that it is not a problem of maximum urls per hosts and also not a problem of the url filter. Looks like the first job of the Generator (Selector) already got to less entries to process. May be this is somehow releasted to split generation or configuration inside the distributed jobtracker since it runs in a different jvm as the jobclient. However we was not able to find the source for this problem. I think that should be fixed before publishing a nutch 0.8. -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: http://issues.apache.org/jira/secure/Administrators.jspa - For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] Created: (NUTCH-195) RPC call times out while indexing map task is computing splits
RPC call times out while indexing map task is computing splits -- Key: NUTCH-195 URL: http://issues.apache.org/jira/browse/NUTCH-195 Project: Nutch Type: Bug Components: indexer Versions: 0.8-dev Environment: MapReduce multi-computer crawl environment: 11 machines (1 master with JobTracker/NameNode, 10 slaves with TaskTrackers/DataNodes) Reporter: Chris Schneider We've been using Nutch 0.8 (MapReduce) to perform some internet crawling. Things seemed to be going well until... 060129 222409 Lost tracker 'tracker_56288' 060129 222409 Task 'task_m_10gs5f' has been lost. 060129 222409 Task 'task_m_10qhzr' has been lost. 060129 222409 Task 'task_r_zggbwu' has been lost. 060129 222409 Task 'task_r_zh8dao' has been lost. 060129 222455 Server handler 8 on 8010 caught: java.net.SocketException: Socket closed java.net.SocketException: Socket closed at java.net.SocketOutputStream.socketWrite(SocketOutputStream.java:99) at java.net.SocketOutputStream.write(SocketOutputStream.java:136) at java.io.BufferedOutputStream.flushBuffer(BufferedOutputStream.java:65) at java.io.BufferedOutputStream.flush(BufferedOutputStream.java:123) at java.io.DataOutputStream.flush(DataOutputStream.java:106) at org.apache.nutch.ipc.Server$Handler.run(Server.java:216) 060129 222455 Adding task 'task_m_cia5po' to set for tracker 'tracker_56288' 060129 223711 Adding task 'task_m_ffv59i' to set for tracker 'tracker_25647' I'm hoping that someone could explain why task_m_cia5po got added to tracker_56288 after this tracker was lost. The Crawl .main process died with the following output: 060129 221129 Indexer: adding segment: /user/crawler/crawl-20060129091444/segments/20060129200246 Exception in thread main java.io.IOException: timed out waiting for response at org.apache.nutch.ipc.Client.call(Client.java:296) at org.apache.nutch.ipc.RPC$Invoker.invoke(RPC.java:127) at $Proxy1.submitJob(Unknown Source) at org.apache.nutch.mapred.JobClient.submitJob(JobClient.java:259) at org.apache.nutch.mapred.JobClient.runJob(JobClient.java:288) at org.apache.nutch.indexer.Indexer.index(Indexer.java:263) at org.apache.nutch.crawl.Crawl.main(Crawl.java:127) However, it definitely seems as if the JobTracker is still waiting for the job to finish (no failed jobs). Doug Cutting's response: The bug here is that the RPC call times out while the map task is computing splits. The fix is that the job tracker should not compute splits until after it has returned from the submitJob RPC. Please submit a bug in Jira to help remind us to fix this. -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: http://issues.apache.org/jira/secure/Administrators.jspa - For more information on JIRA, see: http://www.atlassian.com/software/jira