字符串键不会从 HashMap 中删除

Question

I have a database with urls and their md5 code stored. I need to check if links from a set of urls already present in db. I have the following piece of code, which is run in multiple threads. Each tread is for specific md5 key, which is the first three digits of url's md5:

String[] urls = new String[linksMap.size()];
String[] md5s = new String[linksMap.size()];
boolean expectingMittcom = false;
boolean hadMittcom = false;
int i = 0;
for (String url : linksMap.keySet()) {
  urls[i] = url;
  if (url.equals("http://mittcom.com/")) {
    expectingMittcom = true;
  }
  md5s[i++] = linksMap.get(url).variationMd5;
}

int offset = 0;
while (offset < urls.length) {

  Array arMd5 = pqm.getConnection().createArrayOf("text",
                Arrays.copyOfRange(md5s, offset, Math.min(offset + MAX_NUM,
                urls.length)));
  Array arUrl = pqm.getConnection().createArrayOf("text",
                Arrays.copyOfRange(urls,
                offset, Math.min(offset + MAX_NUM, urls.length)));

  PreparedStatement ps = pqm.getConnection().prepareStatement(
                         "select url from links.links_" + key
                       + " where md5=any(?) and url=any(?)");

  ps.setArray(1, arMd5);
  ps.setArray(2, arUrl);
  ResultSet rs = ps.executeQuery();

  while (rs.next()) {
    String url = rs.getString(1);
    boolean printDebug = false;
    if (url.equals("http://mittcom.com/")) {
      hadMittcom = true;
      printDebug = true;
    }
    LinkVariation r = linksMap.remove(url);
    if (printDebug) {
      logger.info("Link variation: " + r);
    }
    if (r != null) {
      Map<String, String[]> linksMapOriginal = 
           linksByMD5MapOriginal.get(r.original[INDEX_MD5].substring(0, 3));
      if (printDebug) {
        logger.info("will try to fliter out ["
                    + r.original[INDEX_URL] + "]");
      }
      String[] remove = linksMapOriginal.remove(r.original[INDEX_URL]);
      if (remove != null) {
        if (printDebug) {
          logger.info("Filtered mittcom");
          filtered.incrementAndGet();
          checkStillHere();
        }
      } else {
        if (printDebug) {
          logger.info("Did not filter mittcom");
        }
      }
    }
  }

  rs.close();

  ps.close();

  offset += MAX_NUM;
}
if (expectingMittcom) {
  if (hadMittcom) {
    logger.info("was expecting mittcom and found");
  } else {
    logger.info("was expecting mittcom but didn't find");
  }
}

The problem is that url " http://mittcom.com " (and some other else, I just debug for this in particular) still stays in linksByMD5MapOriginal hashMap. I can see in log file that it was removed and filtered, but after threads finish running it is still there! I do not understand how it can happen! I would suspect problems with different hashCode etc, but keys are plain String, there should be no problems like that. I am really puzzled.

I check it like this after all the treads finish running:

    for (Map.Entry<String, Map<String, String[]>> entrySet : linksByMD5MapOriginal.entrySet()) {
        String key = entrySet.getKey();
        Map<String, String[]> value = entrySet.getValue();
        if (value.containsKey("http://mittcom.com/")) {
            logger.info("STILL HERE in " + key);
        }
    }

The hashMap is initilaized as follows:

    protected Map<String, Map<String, String[]>> linksByMD5MapOriginal = new TreeMap<>();

...

    linksByMD5MapOriginal.put(md5Key, linksByKeyMap = Collections.synchronizedMap(new TreeMap<String, String[]>()));

Here TreeMap is for the easier debugging, it does not have to be ordered. The underlying hashMap is synchronized and there should be no problem concurrently modifying it. Nothing adds anything to hashMap while threads are running. Another strange thing is that I cannot use remote debugger (the programs runs on remote server), because the program hangs eventually if I try to do this, so I forced to debug with log printout. But that not the general issue I am asking here for. The problem is that filtered urls still hang in hashMap!

Sorry if my question seems unclear, I will update my post if there will be any follow up questions. Any help will be greatly appreciated.

UPD: log print out:

[2017-10-04 07:25:57,580] [INFO ] [CheckUnique] [Thread-46229] Link variation: routines.queue.CheckUnique$LinkVariation@3f89fc46
[2017-10-04 07:25:57,580] [INFO ] [CheckUnique] [Thread-46229] will try to fliter out [http://www.mittcom.com/]
[2017-10-04 07:25:57,580] [INFO ] [CheckUnique] [Thread-46229] Filtered mittcom
[2017-10-04 07:25:57,580] [INFO ] [CheckUnique] [Thread-46229] Link variation: null
[2017-10-04 07:25:57,580] [INFO ] [CheckUnique] [Thread-46229] Link variation: null
[2017-10-04 07:25:57,580] [INFO ] [CheckUnique] [Thread-46229] Link variation: null
[2017-10-04 07:25:57,580] [INFO ] [CheckUnique] [Thread-46229] Link variation: null
[2017-10-04 07:25:57,580] [INFO ] [CheckUnique] [Thread-46229] Link variation: null
[2017-10-04 07:25:57,580] [INFO ] [CheckUnique] [Thread-46229] Link variation: null
[2017-10-04 07:25:57,580] [INFO ] [CheckUnique] [Thread-46229] Link variation: null
[2017-10-04 07:25:57,580] [INFO ] [CheckUnique] [Thread-46229] Link variation: null
[2017-10-04 07:25:57,580] [INFO ] [CheckUnique] [Thread-46229] Link variation: null
[2017-10-04 07:25:57,580] [INFO ] [CheckUnique] [Thread-46229] was expecting mittcom and found

...

[2017-10-04 07:46:35,337] [INFO ] [CheckUnique] [main] STILL HERE in cd2

Answer 1

I fixed the error and it was somewhere in a piece of code unrelated to this. In short link variation checking mechanism was broken. Can I delete this question?