线上问题整理-ConcurrentModificationException异常

项目场景：

商品改价：商品改价中通过多线程批量处理经过 Lists.partition拆分的集合对象

---

问题描述

商品改价中通过多线程批量处理经过 Lists.partition拆分的集合对象，发现偶尔会报

java.util.ConcurrentModificationException: null
        at java.util.ArrayList$SubList.checkForComodification(ArrayList.java:1231)
        at java.util.ArrayList$SubList.spliterator(ArrayList.java:1235)
        at java.util.Collection.stream(Collection.java:581)

代码实现逻辑

 for (Map.Entry<Integer, List<PriceHandle>> partnerIdProductId : priceHandleGroup.entrySet()) {
            Integer partnerId = partnerIdProductId.getKey();
            List<PriceHandle> priceHandles = partnerIdProductId.getValue();
            log.info("当前分片={},开始同步partnerId={} size={}", shardingItem, partnerId, priceHandles.size());
            List<List<PriceHandle>> pidsGroup = Lists.partition(priceHandles, splitSize);
            log.info("partnerId={},分为{}组，每组{}条", partnerId, pidsGroup.size(), splitSize);
            CountDownLatch countDownLatch = new CountDownLatch(pidsGroup.size());
            for (List<PriceHandle> priceHandlesLittle : pidsGroup) {
                executorService.submit(() -> {
                    try {
                        List<Long> productIds = priceHandlesLittle.stream().map(PriceHandle::getProductId).collect(Collectors.toList());
                        List<ProductMapEntity> productMapEntities = productMappingAllPartnerService.getProductMapListByPartnerIdAndProductIds(partnerId, productIds);
                        //没有映射的失败不处理
                        List<Long> mapProductIds = productMapEntities.stream().map(ProductMapEntity::getProductId).collect(Collectors.toList());
                        List<PriceHandle> noMapPriceHandles = priceHandlesLittle.stream().filter(e -> !mapProductIds.contains(e.getProductId())
                        ).collect(Collectors.toList());
                        if (CollectionUtils.isNotEmpty(noMapPriceHandles)) {
                            jingdongPriceSyncService.updateBatch(noMapPriceHandles, EventStatus.HANDLED_NO_NEED.eventStatus, "商品无映射");
                        }
                        priceHandlesLittle.removeAll(noMapPriceHandles);

                        // 过滤productPool中不存在的重复品
                        filterRepeatProductId(productMapEntities, priceHandlesLittle);
                        if (CollectionUtils.isEmpty(productMapEntities) || CollectionUtils.isEmpty(priceHandlesLittle)) {
                            log.info("过滤productPool中不存在的重复品");
                            return;
                        }
                        PriceSyncService.synPrice(partnerId, productMapEntities, false, priceHandlesLittle);
                    } catch (Exception e) {
                        log.error("同步价格异常,当前处理的店铺：{}", partnerId, e);
                    } finally {
                        countDownLatch.countDown();
                    }
                });
            }
            try {
                countDownLatch.await();
            } catch (InterruptedException e) {
                log.error("countDownLatch.await error", e);
                Thread.currentThread().interrupt();
            }

        }

---

原因分析：

这里需要注意的是在使用Google Guava库中的Lists.partition方法时，如果对原始列表进行了修改，则可能会导致ConcurrentModificationException异常的抛出。这是因为Lists.partition方法返回的是原始列表的视图，而不是副本。因此，如果在对视图进行操作时修改了原始列表，则会导致ConcurrentModificationException异常的抛出。
因此在我们的这段代码中 问题在于又通过拆分后的subList做了removeAll的操作

---

具体复现

解决方案：

方法1不要再for循环中基于视图 修改修改原始列表
方法2 在for循环中新建一个list对象来包装原来的list
具体实现代码

for (Map.Entry<Integer, List<PriceHandle>> partnerIdProductId : priceHandleGroup.entrySet()) {
            Integer partnerId = partnerIdProductId.getKey();
            List<PriceHandle> priceHandles = partnerIdProductId.getValue();
            log.info("当前分片={},开始同步partnerId={} size={}", shardingItem, partnerId, priceHandles.size());
            List<List<PriceHandle>> pidsGroup = Lists.partition(priceHandles, splitSize);
            log.info("partnerId={},分为{}组，每组{}条", partnerId, pidsGroup.size(), splitSize);
            CountDownLatch countDownLatch = new CountDownLatch(pidsGroup.size());
            for (List<PriceHandle> priceHandlesLittle : pidsGroup) {
                // 由于存在并发remove，需要重新赋值新ArrayList，防止ArrayList$SubList引起的ConcurrentModificationException
                List<PriceHandle> copyPriceHandlesLittle = new ArrayList<>(priceHandlesLittle);
                executorService.submit(() -> {
                    try {
                        List<Long> productIds = copyPriceHandlesLittle.stream().map(PriceHandle::getProductId).collect(Collectors.toList());
                        List<ProductMapEntity> productMapEntities = productMappingAllPartnerService.getProductMapListByPartnerIdAndProductIds(partnerId, productIds);
                        //没有映射的失败不处理
                        List<Long> mapProductIds = productMapEntities.stream().map(ProductMapEntity::getProductId).collect(Collectors.toList());
                        List<PriceHandle> noMapPriceHandles = copyPriceHandlesLittle.stream().filter(e -> !mapProductIds.contains(e.getProductId())
                        ).collect(Collectors.toList());
                        if (CollectionUtils.isNotEmpty(noMapPriceHandles)) {
                            PriceSyncService.updateBatch(noMapPriceHandles, EventStatus.HANDLED_NO_NEED.eventStatus, "商品无映射");
                        }
                        copyPriceHandlesLittle.removeAll(noMapPriceHandles);

                        // 过滤productPool中不存在的重复品
                        filterRepeatProductId(productMapEntities, copyPriceHandlesLittle);
                        if (CollectionUtils.isEmpty(productMapEntities) || CollectionUtils.isEmpty(copyPriceHandlesLittle)) {
                            log.info("过滤productPool中不存在的重复品");
                            return;
                        }
                        PriceSyncService.synJPrice(partnerId, productMapEntities, false, copyPriceHandlesLittle);
                    } catch (Exception e) {
                        log.error("同步价格异常,当前处理的店铺：{}", partnerId, e);
                    } finally {
                        countDownLatch.countDown();
                    }
                });
            }
            try {
                countDownLatch.await();
            } catch (InterruptedException e) {
                log.error("countDownLatch.await error", e);
                Thread.currentThread().interrupt();
            }

        }

参考
https://blog.csdn.net/weixin_48321825/article/details/121012733?utm_medium=distribute.pc_relevant.none-task-blog-2_defaultbaidujs_baidulandingword~default-0-121012733-blog-102456357.235^v38pc_relevant_default_base3&spm=1001.2101.3001.4242.1&utm_relevant_index=3